Tech News - Data

It has not been long since the Logan Paul controversy hit the internet and people criticized YouTube algorithms and complained that they were still seeing recommendations of Logan Paul’s videos, even when it was brought down. Earlier this week, a “Flat Earth Conference” was held at Denver, Colorado where some attendees talked about how Youtube has persuaded them to believe the flat earth theory. In fact, Logan Paul was also one of the conference’s keynote speakers, despite not believing that the Earth is flat. The attendees were interviewed by Daily Beast. In the conference, many participants told Daily Beast that they have come to believe in the Flat Earth theory based on YouTube videos. “It came on autoplay,” said Joshua Swift, a conference attendee. “So I didn’t actively search for Flat Earth. Even months before, I was listening to Alex Jones.” Recently, NBA star Kyrie Irving also spoke about his obsession with flat earth theory blaming YouTube videos for it. Irving spoke of having wandered deep down a “rabbit hole” on YouTube. This has brought the emphasis back on the recommendation system that YouTube uses. In a blog post, Guillaume Chaslot, and ex-googler who helped build the YouTube algorithm explains, “Flat Earth is not a ’small bug’. It reveals that there is a structural problem in Google's AIs and they exploit weaknesses of the most vulnerable people, to make them believe the darnedest things.” He mentions a list of Flat Earth videos which were promoted on Youtube. https://www.youtube.com/watch?v=1McqA9ChCnA   https://www.youtube.com/watch?v=XFSH5fnqda4 This makes one question whether the YouTube algorithm is evil? The YouTube algorithm recommends videos based on watch time. More watch time means more revenue and more scope for targeted ads. What this changes, is the fundamental concept of choice and the exercising of user discretion. The moment the YouTube Algorithm considers watch time as the most important metric to recommend videos to you, less importance goes into the organic interactions on YouTube, which includes liking, commenting and subscribing to videos and channels. Chaslot was fired by Google in 2013 over performance issues. His claim was that he wanted to bring about a change in the approach of the YouTube algorithm to make it more aligned with democratic values instead of being devoted to just increasing the watch time. Chaslot has created Algotransparency, a site that scans and monitors YouTube recommendations daily. Other Twitter users have also supported Chaslot’s article. https://twitter.com/tristanharris/status/1064973499540869121 https://twitter.com/technollama/status/1064573492329365504 https://twitter.com/sivavaid/status/1064527872667369473 Is YouTube’s AI Algorithm evil? YouTube has a $25 million plan to counter fake news and misinformation YouTube went down, Twitter flooded with deep questions, YouTube back and everyone is back to watching cat videos

Just two days ago the research team at OpenAI developed Sparse Transformer, a deep neural network that sets new records at predicting what comes next in a sequence, be it text, images, or sound. This transformer uses an algorithmic improvement of the attention mechanism for extracting patterns from sequences that are 30 times longer. This Transformer incorporates an O(N \sqrt{N}) reformulation of the O(N^2) Transformer self-attention mechanism with several other improvements on rich data types. Initially, the models used on these data were designed for one domain. Also, it was difficult to scale to sequences more than a few thousand elements long. The new Sparse Transformer can model sequences with tens of thousands of elements with hundreds of layers for achieving state-of-the-art performance across multiple domains. With this technique, the researchers aim to build AI systems that possess a greater ability to understand the world. The team also introduced several other changes to the Transformer which includes a restructured residual block and weight initialization for improving the training of very deep networks.The team also introduced a set of sparse attention kernels that efficiently compute subsets of the attention matrix. The team further experimented on recomputation of attention weights during the backward pass to reduce memory usage. Initial Experimentation with Deep Attention In Transformers, ‘attention’ is defined as a process where every output element is connected to every input element, and the weightings between them are dynamically calculated based upon the circumstances. Transformers are more flexible than models with fixed connectivity patterns. These Transformers can consume large amounts of memory while being applied to data types with many elements, like images or raw audio. One way of reducing this memory consumption is by recomputing the attention matrix from checkpoints during backpropagation which is a well-established technique in deep learning for reducing memory usage. However, the major issue with recomputing the attention matrix was that it was reducing memory usage at the cost of more computation and also, it couldn’t deal with large inputs. To overcome this, the OpenAI researchers introduced Sparse Attention. Using Sparse Attention patterns for large inputs For very large inputs, computing a single attention matrix can become impractical. The OpenAI researchers instead opted for sparse attention patterns, where each of the output position computes weightings from a subset of input positions. In the entire process, the researchers first visualized the learned attention patterns for deep Transformers on images and then found out that many showed interpretable and structured sparsity patterns. The team also realized that the input portions are focused on small subsets and they show a high degree of regularity. The researchers also implemented a two-dimensional factorization of the attention matrix, where the network can attend to all positions through two steps of sparse attention. They implemented it to preserve the ability of their network to learn new patterns. The first version is strided attention which is roughly equivalent to each position attending to its row and its column and is a bit similar to the attention pattern. The second version is fixed attention which attends to a fixed column and the elements after the latest column element. According to the researchers, it is a useful pattern and can be used when the data doesn’t fit into a two-dimensional structure. Testing Sparse Transformers on density modeling tasks The researchers test their architecture on density modeling tasks including natural images, text, and raw audio using CIFAR-10, Enwik8, and Imagenet 64 datasets respectively.. The team trained strided Sparse Transformers on CIFAR-10 images represented as sequences of 3072 bytes. They also trained models on the EnWik8 dataset for representing the first 108 bytes of Wikipedia containing variability in the periodic structure. They further trained on the version of downsampled ImageNet 64. The researchers found out that sparse attention achieved lower loss than full attention and it is also faster. Future scope and limitations According to the researchers, the sparse attention patterns are only preliminary steps in the direction of efficient modeling of long sequences. The researchers think that exploring different patterns and combinations of sparsity is useful and learning sparse patterns is a promising avenue of research for the next generation of neural network architectures. According to them, the autoregressive sequence generation still seems impractical for very high-resolution images or video. The optimized attention operations may prove to be useful for modeling high dimensional data, like multi-scale approaches. This is just an overview of the Sparse Transformer architecture. For more detailed information, we recommend you to read the research paper. OpenAI Five bots destroyed human Dota 2 players this weekend OpenAI Five beats pro Dota 2 players; wins 2-1 against the gamers OpenAI introduces Neural MMO, a multiagent game environment for reinforcement learning agents

What is Big Data as a Service (BDaaS)? Thanks to the increased adoption of cloud infrastructures, processing, storing, and analyzing huge amounts of data has never been easier. The big data revolution may have already happened, but it’s Big Data as a service, or BDaas, that’s making it a reality for many businesses and organizations. Essentially, BDaas is any service that involves managing or running big data on the cloud. The advantages of BDaas There are many advantages to using a BDaaS solution. It makes many of the aspects that managing a big data infrastructure yourself so much easier. One of the biggest advantages is that it makes managing large quantities of data possible for medium-sized businesses. Not only can it be technically and physically challenging, it can also be expensive. With BDaaS solutions that run in the cloud, companies don’t need to stump up cash up front, and operational expenses on hardware can be kept to a minimum. With cloud computing, your infrastructure requirements are fixed at a monthly or annual cost. However, it’s not just about storage and cos. BDaaS solutions sometimes offer in-built solutions for artificial intelligence and analytics, which means you can accomplish some pretty impressive results without having to have a huge team of data analysts, scientists and architects around you. The different models of BDaaS There are three different BDaaS models. These closely align with the 3 models of cloud infrastructure: IaaS, PaaS, and SaaS. Big Data Infrastructure as a Service (IaaS) – Basic data services from a cloud service provider. Big Data Platform as a Service (PaaS) – Offerings of an all-round Big Data stack like those provided by Amazon S3, EMR or RedShift. This excludes ETL and BI. Big Data Software as a Service (SaaS) – A complete Big Data stack within a single tool. How does the Big Data IaaS Model work? A good example of the IaaS model is Amazon’s AWS IaaS architecture, which combines S3 and EC2. Here, S3 acts as a data lake that can store infinite amounts of structured as well as unstructured data. EC2 acts a compute layer that can be used to implement a data service of your choice and connects to the S3 data. For the data layer you have the option of choosing from among: Hadoop – The Hadoop ecosystem can be run on an EC2 instance giving you complete control NoSQL Databases – These include MongoDB or Cassandra Relational Databases – These include PostgreSQL or MySQL For the compute layer, you can choose from among: Self-built ETL scripts that run on EC2 instances Commercial ETL tools that can run on Amazon’s infrastructure and use S3 Open source processing tools that run on AWS instances, like Kafka How does the Big Data PaaS Model work? A standard Hadoop cloud-based Big Data Infrastructure on Amazon contains the following: Data Ingestion – Logs file data from any data source Amazon S3 Data Storage Layer Amazon EMR – A scalable set of instances that run Map/Reduce against the S3 data. Amazon RDS – A hosted MySQL database that stores the results from Map/Reduce computations. Analytics and Visualization – Using an in-house BI tool. A similar set up can be replicated using Microsoft’s Azure HDInsight. The data ingestion can be made easier with Azure Data Factory’s copy data tool. Apart from that, Azure offers several storage options like Data lake storage and Blob storage that you can use to store results from the computations. How does the Big Data SaaS model work? A fully hosted Big Data stack complete that includes everything from data storage to data visualization contains the following: Data Layer – Data needs to be pulled into a basic SQL database. An automated data warehouse does this efficiently Integration Layer – Pulls the data from the SQL database into a flexible modeling layer Processing Layer – Prepares the data based on the custom business requirements and logic provided by the user Analytics and BI Layer – Fully featured BI abilities which include visualizations, dashboards, and charts, etc. Azure Data Warehouse and AWS Redshift are the popular SaaS options that offer a complete data warehouse solution in the cloud. Their stack integrates all the four layers and is designed to be highly scalable. Google’s BigQuery is another contender that’s great for generating meaningful insights at an unmatched price-performance. Choosing the right BDaaS provider It sounds obvious, but choosing the right BDaaS provider is ultimately all about finding the solution that best suits your needs. There are a number of important factors to consider, such as workload, performance, and cost, each of which will have varying degrees of importance for you. criteria behind the classification include workload, performance and budget requirements. Here are 3 ways you might approach a BDaaS solution:Core BDaaS Core BDaaS uses a minimal platform like Hadoop with YARN and HDFS and other services like Hive. This service has gained popularity among companies which use this for any irregular workloads or as part of their larger infrastructure. They might not be as performance intensive as the other two categories. A prime example would be Elastic MapReduce or EMR provided by AWS. This integrates freely with NoSQL store, S3 Storage, DynamoDB and similar services. Given its generic nature, EMR allows a company to combine it with other services which can result in simple data pipelines to a complete infrastructure. Performance BDaaS Performance BDaaS assists businesses that are already employing a cluster-computing framework like Hadoop to further optimize their infrastructure as well as the cluster performance. Performance BDaaS is a good fit for companies that are rapidly expanding and do not wish to be burdened by having to build a data architecture and a SaaS layer. The benefit of outsourcing the infrastructure and platform is that companies can focus on specific processes that add value instead of concentrating on complicated Big Data related infrastructure. For instance, there are many third-party solutions built on top of Amazon or Azure stack that let you outsource your infrastructure and platform requirements to them. Feature BDaaS If your business is in need of additional features that may not be within the scope of Hadoop, Feature BDaaS may be the way forward. Feature BDaaS focuses on productivity as well as abstraction. It is designed to enable users to be up and using Big Data quickly and efficiently. Feature BDaaS combines both PaaS and SaaS layers. This includes web/API interfaces, and database adapters that offer a layer of abstraction from the underlying details. Businesses don’t have to spend resources and manpower setting up the cloud infrastructure. Instead, they can rely on third-party vendors like Qubole and Altiscale that are designed to set it up and running on AWS, Azure or cloud vendor of choice quickly and efficiently. Additional Tips for Choosing a Provider When evaluating a BDaaS provider for your business, cost reduction and scalability are important factors. Here are a few tips that should help you choose the right provider. Low or Zero Startup Costs – A number of BDaaS providers offer a free trial period. Therefore, theoretically, you can start seeing results before you even commit a dollar. Scalable – Growth in scale is in the very nature of a Big Data project. The solution should be easy and affordable to scale, especially in terms of storage and processing resources. Industry Footprint – It is a good idea to choose a BDaaS provider that already has an experience in your industry. This is doubly important if you are also using them for consultancy and project planning requirements. Real-Time Analysis and Feedback – The most successful Big Data projects today are those that can provide almost immediate analysis and feedback. This helps businesses to take remedial action instantly instead of working off of historical data. Managed or Self-Service – Most BDaaS providers today provide a mix of both managed as well as self-service models based on the company’s needs. It is common to find a host of technical staff working in the background to provide the client with services as needed. Conclusion The value of big data is not in the data itself, but in the insights that can be drawn after processing it and running it through robust analytics. This can help to guide and define your decision making for the future. A quick tip with regards to using Big Data: keep it small at the initial stages. This ensures the data can be checked for accuracy and the metrics derived from them are right. Once confirmed, you can go ahead with more complex and larger data projects. Gilad David Maayan is a technology writer who has worked with over 150 technology companies including SAP, Oracle, Zend, CheckPoint and Ixia. Gilad is a 3-time winner of international technical communication awards, including the STC Trans-European Merit Award and the STC Silicon Valley Award of Excellence. Over the past 7 years Gilad has headed Agile SEO, which performs strategic search marketing for leading technology brands. Together with his team, Gilad has done market research, developer relations and content strategy in 39 technology markets, lending him a broad perspective on trends, approaches and ecosystems across the tech industry. Common big data design patterns Hortonworks partner with Google Cloud to enhance their Big Data strategy Top 5 programming languages for crunching Big Data effectively Getting to know different Big data Characteristics  

Amanda Boyle Tanna Solberg December 8, 2020 - 12:25am December 8, 2020 In their mastery of the Tableau platform, their desire to collaborate and help invent the Tableau solutions of tomorrow, and their dedication to helping our global community, Tableau Zen Masters stand out in a community of greatness. They are also our biggest advocates as well as our harshest critics, but by listening to and engaging with the Tableau Community, we build better software and, ultimately, a better company. The spirit of Tableau is our customers. And as one way to better support our customers, we are growing the Zen Master program this year. We are looking to add more diverse leaders and to grow representation throughout the world. We are excited to announce that it’s time to nominate a new cohort of Zen Masters. This means we need your help! To ensure a diverse pool of nominees representative of our global community, we need the collective force of all of you to help us spread the news to your colleagues and your friends. We need you to champion your peers, showcase, and shine a light on those doing incredible work that elevates others in a public space. By nominating a fellow community member to become a Tableau Zen Master, you are not only recognizing your data heroes, but you are also giving your input about who you want to lead the Tableau Community for the upcoming year. As we shared in the 2020 Ambassador nomination process, we are listening to the calls for action and change from the community—and from our own team. We ask you to elevate diverse voices by nominating Black, Indigenous, and people of color to lead our community. We know that by amplifying diverse leaders in our community, we can help create better outcomes for all, and better reflect the communities in which we live and work. To establish greater equity, we need to bring diversity into this community proactively. The Tableau Community Equity Task Force will be advising our team on recruitment opportunities, but it will take a collective effort to be successful. Thank you for your support and engagement. We avoid selecting individuals whose work is internal and private to their organization. We respect that as our communities have grown, internal communities have flourished. Undoubtedly, we want to celebrate these efforts, but continue to only select members whose work meets our criteria of working and elevating others in public spaces—and not fee-gated. What makes someone a Tableau Zen Master? The Zen Master nomination period begins now, and will be open from Tuesday, December 8, 2020, through Friday, January 8, 2021. During the nomination period, we invite you to highlight the people who inspire and instruct you—those with exceptional dedication to exploring Tableau and improving it. The “humble-smart” leaders who make the community so remarkable. Zen Master selections are made from nominations and applications from you, members of the Tableau Community! When submitting a nomination, you will be asked to share examples of how you or your nominee has demonstrated the three pillars of the Tableau Zen Master program: teacher, master, and collaborator. As you prepare, consider the following questions: How has this person served as a teacher in the last year? Does the person dedicate their time to helping others be better at using Tableau? Are they a Tableau evangelist who shares our mission of helping people see and understand data? Does the person add to the Tableau Community by offering help to people of all levels? How do you or your nominee demonstrate mastery on the Tableau platform? Has the person shown a deep understanding of how Tableau works? They might create beautiful dashboards on Tableau Public, maintain Tableau Server, build mind-blowing extensions, or more. How does your nominee collaborate? Does the person work with others to bring Tableau and its mission to new communities and geographies? Have they worked with other community members to build thought leadership or training resources? Do they contribute useful, popular ideas on our Ideas Forum? If all of these attributes can be found in someone you know, nominate them to be a Tableau Zen Master. Please be brief and focused in your response, including links to blogs, Tableau Public profiles, vizzes, virtual event links, and other sources. Tableau and Salesforce employees, partners, and community members are all welcome to submit nominations. Ready, set, nominate! Please complete one submission for each person you want to nominate. Nominations close at 10:00 pm PST on Friday, January 8, 2021. All nominations will be reviewed by a selection committee made up of Tableau employees with input from the Hall of Fame Zen Masters. We do not select Zen Masters based on the number of nominations received. While we do read and track all nominations, we also use internal metrics, employee nominations, and the needs of our global community to determine the new cohort. Further details can be found on the Tableau Zen Master page. Getting together looked different this year: Tableau Community members including Zen Masters, Ambassadors & friends coming together for TC’ish Supporting our Community through 2020 and beyond In February 2020, we invited 34 individuals, representing 11 countries, and 4 global regions to serve as Tableau Zen Masters for a one-year term.  This year’s Zen Masters helped design and pivot events to virtual platforms—welcoming thousands to the #DataFamCommunityJams. They supported new mentorship initiatives to help people feel more connected in isolation and build new opportunities for collaboration. They worked countless hours standing up the first iteration of what would become the Tableau COVID-19 Data Hub. These leaders jumped in without hesitation when requests came in from global public-health leaders desperate for assistance with organizing and analyzing data. The 2020 Zen Masters brought their passion and expertise to a new generation, creating content for our Data Literacy for All eLearning program that provides data skills fundamentals, regardless of skill level. And just last month, two Hall of Fame Zen Masters gave their time to work with SurveyMonkey and Axios to make sure we put out the best possible and best-performing visualizations in our US Election Polling partnership for the presidential election.  Zen Masters Jeffrey Shaffer, Sarah Bartlett, and Kevin Flerlage joined by Ambassador Adam Mico, Dinushki De Livera, and Mark Bradbourne at the Cincinnati TUG in January, 2020 We are inviting all 2020 Tableau Zen Masters to join us for another year. 2020 didn't quite work out the way anyone predicted. The pressures of COVID-19, combined with so many other factors have had an impact on everyone personally and professionally—and have also impacted the Zen Master experience. We encouraged all of our community leaders to prioritize their health and wellness, and that of their families. We supported members disengaging to take care of more pressing priorities, and we greatly appreciate that they did. Through it all, the 2020 class exceeded all expectations as teachers, masters, and collaborators in brilliant, meaningful ways that truly embodied the pay-it-forward spirit of the Tableau Community.  We have offered the 2020 Zen Masters an early invitation to join the 2021 group. There are a few reasons why we have made this decision. First, this year’s group has had unique, pandemic-related challenges in terms of access to Tableau teams, speaking opportunities, and time to connect with one another, as well as a lack of support from our team. Second, we just think it’s the right thing to do. We know this year has been challenging. We are learning as we go, and we want all the current Zen Masters to have a meaningful experience—one we believe we have not provided this year. This will not be an extension of the current year and will add to the 5-year minimum to be considered for the Zen Master Hall of Fame. Current Zen Masters are being asked to provide a recap of their experience, sharing what is or is not working for them, and any feedback to help strengthen the program through the next year.  Left: Zen Master Ann Jackson sharing her knowledge and passion for teaching and problem solving as a volunteer Tableau Doctor at TC’ish 2020, Right: Zen Master Yukari Nagata supporting the APAC CommunityAll Zen Masters completing their 5th term will be eligible for 2021 Hall of Fame consideration. Current Zen Masters who will be eligible after completing their 5th term include Adam McCann, Chris Love, Jeffrey Shaffer, Rob Radburn, and Tamas Foldi. Each member will go through a similar evaluation process that we have used with previous groups. Thank you, #datafam, for being a part of the Tableau Community! We look forward to hearing from you.

A California-based AI startup, Cerebras Systems has unveiled the largest semiconductor chip ever built named as the ‘Wafer Scale Engine’ built to quickly train deep learning models. The Cerebras Wafer Scale Engine (WSE) is 46,225 millimeters square, contains more than 1.2 trillion transistors. It is “more than 56X larger than the largest graphics processing unit, containing 3,000X more on-chip memory and more than 10,000X the memory bandwidth,” the whitepaper reads. Most of the chips available today include a collection of chips built on top of a 12-inch silicon wafer and are processed in a chip factory in a batch.  However, the WSE chip is interconnected on a single wafer. “The interconnections are designed to keep it all functioning at high speeds so the trillion transistors all work together as one,” Venture Beats reports. Andrew Feldman, co-founder and CEO of Cerebras system said, “Designed from the ground up for AI work, the Cerebras WSE contains fundamental innovations that advance the state-of-the-art by solving decades-old technical challenges that limited chip size — such as cross-reticle connectivity, yield, power delivery, and packaging.” He further adds, “Every architectural decision was made to optimize performance for AI work. The result is that the Cerebras WSE delivers, depending on workload, hundreds or thousands of times the performance of existing solutions at a tiny fraction of the power draw and space.” According to Wired, “Cerebras’ chip covers more than 56 times the area of Nvidia’s most powerful server GPU, claimed at launch in 2017 to be the most complex chip ever. Cerebras founder and CEO Andrew Feldman says the giant processor can do the work of a cluster of hundreds of GPUs, depending on the task at hand, while consuming much less energy and space.” Source: Twitter In the whitepaper, Feldman explains, for maximum performance, the entire model should fit in the fastest memory, which is the memory closest to the computation cores. This is not the case in CPUs, TPUs, and GPUs, where main memory is not integrated with compute. Instead, the vast majority of memory is based off-chip, far away on separate DRAM chips or a stack of these chips in a high bandwidth memory (HBM) device. As a result, the main memory is excruciatingly slow. The dawn of AI brought in an added consumption of higher processing power which gave rise to the demand of GPUs. However, even if a machine is filled with dozens of Nvidia’s graphics chips or GPUs, “it can take weeks to “train” a neural network, the process of tuning the code so that it finds a solution to a given problem,” according to Fortune. Linley Gwennap, a chip observer who publishes a distinguished chip newsletter, Microprocessor Report told Fortune that bundling together multiple GPUs in a computer starts to show diminishing returns once more than eight of the chips are combined. Feldman further adds “The hard part is moving data.” While training a neural network, thousands of operations happen in parallel. Also, chips must constantly share data as they crunch those parallel operations. However, computers with multiple chips may face performance issues while trying to pass data back and forth between the chips over the slower wires that link them on a circuit board. The solution Feldman said was to “take the biggest wafer you can find and cut the biggest chip out of it that you can.” Per Fortune, “the chip won’t be sold on its own but will be packaged into a computer “appliance” that Cerebras has designed. One reason is the need for a complex system of water-cooling, a kind of irrigation network to counteract the extreme heat generated by a chip running at 15 kilowatts of power.” “The wafers were produced in partnership with Taiwan Semiconductor Manufacturing, the world’s largest chip manufacturer, but Cerebras has exclusive rights to the intellectual property that makes the process possible.” J.K. Wang, TSMC’s senior vice president of operations said, “We are very pleased with the result of our collaboration with Cerebras Systems in manufacturing the Cerebras Wafer Scale Engine, an industry milestone for wafer-scale development.” “TSMC’s manufacturing excellence and rigorous attention to quality enable us to meet the stringent defect density requirements to support the unprecedented die size of Cerebras’ innovative design.” The whitepaper explains that 400,000 cores on Cerebras WSE are connected via a Swarm communication fabric in a 2D mesh with 100 Petabits per second of bandwidth. Swarm provides a hardware routing engine to each of the compute cores and connects them with short wires optimized for latency and bandwidth. Feldman said that “a handful” of customers are trying the chip, including on drug design problems. He plans to sell complete servers built around the chip, rather than chips on their own  but declined to discuss price or availability. Many find this announcement interesting given the number of transistors in work on the wafer engine. A few are skeptical if this chip will live up to the expectation. A user on Reddit commented, “I think this is fascinating. If things go well with node scaling and on-chip non-volatile memory, by mid 2030 we could be approaching human brain densities on a single ‘chip’ without even going 3D. It's incredible.” A user on HackerNews writes, “In their whitepaper, they claim "with all model parameters in on-chip memory, all of the time," yet that entire 15 kW monster has only 18 GB of memory. Given the memory vs compute numbers that you see in Nvidia cards, this seems strangely low.” https://twitter.com/jwangARK/status/1163928272134168581 https://twitter.com/jwangARK/status/1163928655145426945 To know more about Cerebras WSE chip in detail, read the complete whitepaper. Why DeepMind AlphaGo Zero is a game changer for AI research Speech2Face: A neural network that “imagines” faces from hearing voices. Is it too soon to worry about ethnic profiling? Alibaba’s chipmaker launches open source RISC-V based ‘XuanTie 910 processor’ for 5G, AI, IoT and self-driving applications

“Debugging code is drudgery” -Facebook Engineers Yue Jia, Ke Mao and Mark Harman To significantly reduce the amount of time developers spend on debugging code and rolling out new software, Facebook engineers have come up with an ingenious tool called ‘SapFix’. Sapfix, which is still under development, can automatically generate fixes for specific bugs  identified by Sapienz. It will then propose these fixes to engineers for approval and deployment to production. SapFix will eventually be able to operate independently from Sapienz, Facebook’s intelligent automated software testing tool. For now, it is a proof-of-concept that relies on the latter tool to pinpoint bugs. How does SapFix work? This AI hybrid tool will generate bug fixes depending upon the type of bug encountered. For instance: For simpler bugs: SapFix will create patches that revert the code submission that introduced these bugs. For complicated bugs: The tool uses a collection of “templated fixes” that were created by human engineers based on previous bug fixes. If human-designed template fixes aren’t up to the job: The tool attempts a “mutation-based fix,” which works by continuously making small modifications to the code that caused the software to crash, until a solution is found. SapFix generates multiple potential fixes for every bug. This is then submitted to the engineers for evaluation. The fixes are tested in advance so engineers can check if they might cause problems like compilation errors and other crashes. Source: Facebook With an automated end-to-end testing and repair, SapFix is an important milestone in AI hybrid tool deployment. Facebook intends to open source both, SapFix and Sapienz, once additional engineering work has been completed. You can read more about this tool on Facebook’s Blog. Facebook introduces Rosetta, a scalable OCR system that understands text on images using Faster-RCNN and CNN How AI is going to transform the Data Center Facebook Reality Labs launch SUMO Challenge to improve 3D scene understanding and modeling algorithms  

[box type="note" align="" class="" width=""]In this article by Shilpi Saxena and Saurabh Gupta from their book Practical Real-time data Processing and Analytics we shall explore Storm's architecture with its components and configure it to run in a cluster. [/box] Initially, real-time processing was implemented by pushing messages into a queue and then reading the messages from it using Python or any other language to process them one by one. The primary challenges with this approach were: In case of failure of the processing of any message, it has to be put back in to queue for reprocessing Keeping queues and the worker (processing unit) up and running all the time Below are the two main reasons that make Storm a highly reliable real-time engine: Abstraction: Storm is distributed abstraction in the form of Streams. A Stream can be produced and processed in parallel. Spout can produce new Stream and Bolt is a small unit of processing on stream. Topology is the top level abstraction. The advantage of abstraction here is that nobody must be worried about what is going on internally, like serialization/deserialization, sending/receiving message between different processes, and so on. The user must be focused on writing the business logic. A guaranteed message processing algorithm:  Nathan Marz developed an algorithm based on random numbers and XORs that would only require about 20 bytes to track each spout tuple, regardless of how much processing was triggered downstream. Storm Architecture and Storm components The nimbus node acts as the master node in a Storm cluster. It is responsible for analyzing topology and distributing tasks on different supervisors as per the availability. Also, it monitors failure; in the case that one of the supervisors dies, it then redistributes the tasks among available supervisors. Nimbus node uses Zookeeper to keep track of tasks to maintain the state. In case of Nimbus node failure, it can be restarted which reads the state from Zookeeper and start from the same point where it failed earlier. Supervisors act as slave nodes in the Storm cluster. One or more workers, that is, JVM processes, can run in each supervisor node. A supervisor co-ordinates with workers to complete the tasks assigned by nimbus node. In the case of worker process failure, the supervisor finds available workers to complete the tasks. A worker process is a JVM running in a supervisor node. It has executors. There can be one or more executors in the worker process. Worker co-ordinates with executor to finish up the task. An executor is single thread process spawned by a worker. Each executor is responsible for running one or more tasks. A task is a single unit of work. It performs actual processing on data. It can be either Spout or Bolt. Apart from above processes, there are two important parts of a Storm cluster; they are logging and Storm UI. The logviewer service is used to debug logs for workers at supervisors on Storm UI. The following are the primary characteristics of Storm that make it special and ideal for real-time processing. Fast Reliable Fault-Tolerant Scalable Programming Language Agnostic Strom Components Tuple: It is the basic data structure of Storm. It can hold multiple values and data type of each value can be different. Topology: As mentioned earlier, topology is the highest level of abstraction. It contains the flow of processing including spout and bolts. It is kind of graph computation. Stream: The stream is core abstraction of Storm. It is a sequence of unbounded tuples. A stream can be processed by the different type of bolts and which results into a new stream. Spout: Spout is a source of stream. It reads messages from sources like Kafka, RabbitMQ, and so on as tuples and emits them in a stream. There are two types of Spout Reliable: Spout keeps track of each tuple and replay tuple in case of any failure. Unreliable: Spout does not care about the tuple once it is emitted as a stream to another bolt or spout. Setting up and configuring Storm Before setting up Storm, we need to setup Zookeeper which is required by Storm: Setting up Zookeeper Below are instructions on how to install, configure and run Zookeeper in standalone and cluster mode: Installing Download Zookeeper from http://www-eu.apache.org/dist/zookeeper/zookeeper-3.4.6/zookeeper-3.4.6.tar.gz. After the download, extract zookeeper-3.4.6.tar.gz as below: tar -xvf zookeeper-3.4.6.tar.gz The following files and folders will be extracted: Configuring There are two types of deployment with Zookeeper; they are standalone and cluster. There is no big difference in configuration, just new extra parameters for cluster mode. Standalone As shown, in the previous figure, go to the conf folder and change the zoo.cfg file as follows: tickTime=2000 # Length of single tick in milliseconds. It is used to # regulate heartbeat and timeouts. initLimit=5 # Amount of time to allow followers to connect and sync # with leader. syncLimit=2 # Amount of time to allow followers to sync with # Zookeeper dataDir=/tmp/zookeeper/tmp # Directory where Zookeeper keeps # transaction logs clientPort=2182 # Listening port for client to connect. maxClientCnxns=30 # Maximum limit of client to connect to Zookeeper # node. Cluster In addition to above configuration, add the following configuration to the cluster as well: server.1=zkp-1:2888:3888 server.2=zkp-2:2888:3888 server.3=zkp-3:2888:3888 server.x=[hostname]nnnn:mmmm : Here x is id assigned to each Zookeeper node. In datadir, configured above, create a file "myid" and put corresponding ID of Zookeeper in it. It should be unique across the cluster. The same ID is used as x here. Nnnn is the port used by followers to connect with leader node and mmmm is the port used for leader election. Running Use the following command to run Zookeeper from the Zookeeper home dir: /bin/zkServer.sh start The console will come out after the below message and the process will run in the background. Starting zookeeper ... STARTED The following command can be used to check the status of Zookeeper process: /bin/zkServer.sh status The following output would be in standalone mode: Mode: standalone The following output would be in cluster mode: Mode: follower # in case of follower node Mode: leader # in case of leader node Setting up Apache Storm Below are instructions on how to install, configure and run Storm with nimbus and supervisors. Installing Download Storm from http://www.apache.org/dyn/closer.lua/storm/apache-storm-1.0.3/apache-storm-1.0.3.tar.gz. After the download, extract apache-storm-1.0.3.tar.gz, as follows: tar -xvf apache-storm-1.0.3.tar.gz Below are the files and folders that will be extracted: Configuring As shown, in the previous figure, go to the conf folder and add/edit properties in storm.yaml: Set the Zookeeper hostname in the Storm configuration: storm.zookeeper.servers: - "zkp-1" - "zkp-2" - "zkp-3" Set the Zookeeper port: storm.zookeeper.port: 2182 Set the Nimbus node hostname so that storm supervisor can communicate with it: nimbus.host: "nimbus" Set Storm local data directory to keep small information like conf, jars, and so on: storm.local.dir: "/usr/local/storm/tmp" Set the number of workers that will run on current the supervisor node. It is best practice to use the same number of workers as the number of cores in the machine. supervisor.slots.ports: - 6700 - 6701 - 6702 - 6703 - 6704 - 6705 Perform memory allocation to the worker, supervisor, and nimbus: worker.childopts: "-Xmx1024m" nimbus.childopts: "-XX:+UseConcMarkSweepGC – XX:+UseCMSInitiatingOccupancyOnly – XX_CMSInitiatingOccupancyFraction=70" supervisor.childopts: "-Xmx1024m" Topologies related configuration: The first configuration is to configure the maximum amount of time (in seconds) for a tuple's tree to be acknowledged (fully processed) before it is considered failed. The second configuration is that Debug logs are false, so Storm will generate only info logs. topology.message.timeout.secs: 60 topology.debug: false Running There are four services needed to start a complete Storm cluster: Nimbus: First of all, we need to start Nimbus service in Storm. The following is the command to start it: /bin/storm nimbus Supervisor: Next, we need to start supervisor nodes to connect with the nimbus node. The following is the command: /bin/storm supervisor UI: To start Storm UI, execute the following command: /bin/storm ui You can access UI on http://nimbus-host:8080. It is shown in following figure. Logviewer: Log viewer service helps to see the worker logs in the Storm UI. Execute the following command to start it: /bin/storm logviewer Summary We started with the history of Storm, where we discussed how Nathan Marz the got idea for Storm and what type of challenges he faced while releasing Storm as open source software and then in Apache. We discussed the architecture of Storm and its components. Nimbus, supervisor worker, executors, and tasks are part of Storm's architecture. Its components are tuple, stream, topology, spout, and bolt. We discussed how to set up Storm and configure it to run in the cluster. Zookeeper is required to be set up first, as Storm requires it. The above was an excerpt from the book Practical Real-time data Processing and Analytics 

Reptile, developed by Open AI, is a simple meta-learning algorithm. Meta-learning is the process of learning how to learn. A meta-learning algorithm takes in a distribution of tasks, where each task is a learning problem, and it produces a quick learner. This means a learner must be able to generalize from a small number of examples. An example of a meta-learning problem is few-shot classification. Here, each task is a classification problem within which the learner after seeing only 1 - 5 input-output examples from each class must classify new inputs. What Reptile does It samples a task repeatedly, performs stochastic gradient descent on it, and finally updates the initial parameters towards the final parameters learned on the task. Any Comparisons? Reptile performs as well as MAML, which is also a broadly applicable meta-learning algorithm. Unlike MAML, Reptile is simple to implement and more computationally efficient. Some features of Reptile : Reptile seeks an initialization for the parameters of a neural network, such that the network can be fine-tuned using a small amount of data from a new task. Unlike MAML, Reptile simply performs stochastic gradient descent (SGD) on each task in a standard way. This means it does not unroll a computation graph or calculate any second derivatives. Hence, Reptile takes less computation and memory than MAML. The current Reptile implementation uses TensorFlow for the computations involved, and includes code for replicating the experiments on Omniglot and Mini-ImageNet. To Read more on how Reptile works, visit the OpenAI blog. To view Reptile implementations, visit its GitHub Repository.  

Facebook data scientists had released a paper, Constrained Bayesian Optimization with Noisy Experiments in 2017 where they describe using Bayesian optimization to design rounds of A/B tests based on prior test results. An A/B test is a randomized experiment, used to determine which variant of A and B is more "effective". They are used for improving a product. Facebook has a large array of backend systems serving billions of people every day. They have a large number of internal parameters that must be tuned carefully using live, randomized experiments, also known as A/B tests. Individual experiments may take a week or longer, so there is a challenge to optimize a set of parameters with the least number of experiments. Bayesian optimization Bayesian optimization is a technique used to solve optimization problems where the objective function (the online metric of interest) does not have an analytic expression. It can only be evaluated through some time consuming operations like a randomized experiment. Bayesian optimization allows joint tuning of more parameters with fewer experiments compared to a grid search or manual tuning. It also helps in finding better values. The Gaussian process (GP) is a Bayesian model that works well for Bayesian optimization. GP provides good uncertainty estimates of how an online metric varies with the parameters of interest. It is illustrated as follows: Source: Facebook research blog The work in the paper was motivated by several challenges in using Bayesian optimization for tuning online systems. The challenges are noise, constraints, and batch experimentation. In the paper, the authors describe a Bayesian approach for handling observation noise in which they include the posterior uncertainty induced by noise in EI’s expectation. In the paper, they describe a Bayesian approach for handling observation noise. A posterior uncertainty is induced by noise in EI’s expectation. Instead of computing the expectation of I(x) under the posterior of f(x), it is computed under the joint posterior of f(x) and f(x*). This expectation no longer has a closed form like El but can easily draw samples of values at past observations f(x_1), …, f(x_n) from the GP posterior. The conditional distribution f(x) | f(x_1), …, f(x_n) has closed form. The results The approach described in the paper is used to optimize various systems at Facebook. Two such optimizations are described in the paper. The first is to optimize six parameters from one of Facebook’s ranking systems. The second one was to optimize seven numeric compiler flags for the HipHop Virtual Machine (HHVM). The web servers powering Facebook use the HHVM to serve requests. The end goal of this optimization was to reduce CPU usage on the web servers, with a constraint of keeping the peak memory usage less. This following figure shows the CPU usage of each configuration tested. There is a 100 total, it also shows the probability that each point satisfied the memory constraint: Source: Facebook research blog The first 30 iterations were randomly generated configurations depicted as a green line. After this, the Bayesian optimization was used to identify parameter configurations to be evaluated. It was observed that Bayesian optimization was able to find better configurations that are more likely to satisfy the constraints. The findings are that Bayesian optimization is an effective and robust tool for optimizing via noisy experiments. For full details, visit the Facebook research blog. You can also take a look at the research paper. NIPS 2017 Special: A deep dive into Deep Bayesian and Bayesian Deep Learning with Yee Whye Teh Facebook’s Glow, a machine learning compiler, to be supported by Intel, Qualcomm and others “Deep meta reinforcement learning will be the future of AI where we will be so close to achieving artificial general intelligence (AGI)”, Sudharsan Ravichandiran

Give it four days to practice and you would have a chess master ready!  This line stands true for Deepmind’s latest AI program, AlphaZero. AlphaZero is an advanced version of AlphaGo Zero--the AI that recently won all games of Go against its precursor AlphaGo--relies simply on self-play without any example games. AlphaZero is an improvement to it as it shows that the same program can master three different types of board games, Chess, Shogi and Go namely. It uses reinforcement learning algorithm to achieve state-of-the-art results. AlphaZero mastered the game of chess, without having prior domain knowledge of the game, except the game rules. Additionally, it also mastered Shogi, a Japanese board game, as showcased in a recent DeepMind research paper. Demis Hassabis, founder, and CEO, DeepMind introduced some additional details of AlphaZero at the Neural Information Processing Systems (NIPS) conference in Long Beach, California. “It doesn’t play like a human, and it doesn’t play like a program, it plays in a third, almost alien, way,” said Hassabis. It only took four hours to self-play and create chess knowledge beyond any human or computer program. Surprisingly, it defeated Stockfish 8 (A world champion chess engine) in four hours without any external help or any prior empirical data (a database of archived chess games, or well-known chess strategies and openings). The hyper-parameter of AlphaGo Zero’s search was tuned by using Bayesian optimization algorithm. AlphaZero reuses the same hyper-parameter for playing all the board games without performing any game-specific tuning. Similar to AlphaGo Zero, AlphaZero’s board state is encoded by spatial planes based on specifically the basic rules for each game. While training AlphaZero, the same algorithmic settings, network architecture, and hyper-parameters were used in all three games. A separate instance of AlphaZero was trained for each game. The training initiated for 700,000 steps (mini-batches of size 4,096) starting from randomly initialized parameters, with 5,000 first-generation TPUs to generate self-play games and 64 second-generation TPUs to train the neural networks. After comprehensive analysis, it was found that AlphaZero outperformed Stockfish in Chess in 4 hours Elmo in Shogi in less than 2 hrs AlphaGo Lee in Go in 8 hours The achievements by AlphaZero are impressive, to say the least. Researchers at DeepMind say that it still needs to play many more practice games than a human chess champion. Human learning is based on watching other people play and also by learning in different ways, which a machine cannot achieve. But it can go beyond human thinking by expanding the capabilities of its program. To know more about how AlphaZero masters chess and Shogi using Reinforcement algorithm, you can have a look at the research paper here or tune into the game series on Youtube to watch the video.

Yesterday, a shocking report by The New York Times shared its investigation on sexual misconduct at Google. It alleged that Google had protected at least four senior executives over the past decade after they were accused of sexual misconduct. They obtained corporate and court documents and spoke to more than three dozen current and former Google executives and employees about these episodes. Here is a summary of the three incidents that the New York Times article reported on.. The controversy with Andy Rubin, Creator of Android Andy Rubin, the creator of Android, often exhibited unprofessional behavior towards his co-workers. He was involved in a consensual relationship with a woman employee from 2011, who reported to one of his direct reports on the Android team. Google’s human resources department was not informed about this relationship despite a policy in place to do so. In 2013 when she wanted to cool things off, she agreed to meet Rubin at a hotel, where she was pressured to perform a non-consensual sexual activity. The woman filed a complaint to Google’s human resources department in 2014 and informed officials about the relationship. Amidst Google’s investigation, in September 2014, Mr. Rubin was awarded a stock grant worth $150 million approved by Google board’s leadership development and compensation committee. Google’s inquiry found the claims to be credible and the relationship inappropriate.   Mr. Page, the then CEO of Google, decided Mr. Rubin should leave and Google paid him $90 million as an exit package with an agreement to not work with Google’s rival companies. The company then proceeded to give Mr. Rubin’s a high profile well-respected farewell in October 2014. A civil suit filed later this month by Mr. Rubin’s ex-wife, Rie Rubin, includes a screenshot on an email (dated August 2015) that Mr. Rubin sent to a woman which said: “You will be happy being taken care of, Being owned is kinda like you are my property, and I can loan you to other people.” Mr. Rubin released a statement calling the allegations “false” and “part of a smear campaign by my ex-wife to disparage me during a divorce and custody battle.” The controversy with Richard DeVaul, Director at Google X In 2013, Richard DeVaul, director at Google X, interviewed Star Simpson, a hardware engineer. After the job interview, he invited her to an annual festival in the Nevada desert, the following week. On getting back to his encampment, he asked her to remove her shirt and offered a back rub. When she refused, he insisted and she relented to a neck rub. Why you ask? “I didn’t have enough spine or backbone to shut that down as a 24-year-old” -Ms. Simpson Later she was informed by Google that she did not land the job, without any explanation. After finally reporting the episode to Google after 2 years, human resources told her that her account was “more likely than not” true and that “appropriate action” was taken. She was asked to stay quiet about the whole incident. Chelsea Bailey, the head of human resources at X, declined Simpson's allegations in a statement, adding that officials investigated and “took appropriate corrective action.” declining to say what the action was, owing to employee confidentiality. The controversy with Amit Singhal, former SVP of Search In 2005, an employee alleged that Amit Singhal, a senior vice president who headed search, groped her at an off-site event attended by dozens of colleagues. Google investigated and found that Mr. Singhal was inebriated and there were no witnesses to corroborate the incident. Google did not fire Mr. Singhal. They accepted his resignation and negotiated an exit package that paid him millions and prevented him from working for a competitor. The controversy with Drummond, Chief Legal Officer, Alphabet, and Chairman, CapitalG “Google felt like I was the liability.” - Jennifer Blakely, ex- senior contract manager David C. Drummond, joined as general counsel in 2002, started dating Jennifer Blakely (senior contract manager) in 2004. They had a son in 2007, after which Mr. Drummond disclosed their relationship to the company. Soon after, Google took action and Ms. Blakely had to leave the legal department as only one of them could work there and transferred to sales in 2007. She eventually left Google in 2008. While resigning, she was asked to sign paperwork saying she had departed voluntarily. Drummond left her in late 2008. Since the affair, Mr. Drummond’s has moved up the rungs within Alphabet. As Alphabet’s chief legal officer and chairman of CapitalG, he has reaped about $190 million from stock options and awards since 2011. Google’s response to the New York Times story Following the report by the New York Times, Google CEO Sundar Pichai sent an email to all Google employees on Thursday clarifying that the company has fired 48 people over the last two years for sexual harassment 13 of them were "senior managers and above". None of them received any exit packages. The email opened “We are dead serious about making sure we provide a safe and inclusive workplace. We want to assure you that we review every single complaint about sexual harassment or inappropriate conduct, we investigate and we take action.” It also stated that there are “confidential channels” available for employees to report incidents of sexual harassment. He further informed they have updated their policies to demand all VPs and SVPs to disclose any relationship with a co-worker irrespective of whether they work on the same projects or not.  You can head over to CNBC to read the entire email. Our take on this story The email seems to have deliberately excluded the timelines during which the incidents reported in the New York Times article took place. Also, it neither denies nor confirms those incidents which hints at them being true, in most likelihood. While Mr. Pichai assures his people that Google is doing everything to ensure it is a safe place to work, he does not address any of the red flags satisfactorily the NYT article raised such as: All the above incidents point to weak policy implementation by HR and Google leadership. Just amending policies is clearly not enough. The ‘hard line on inappropriate conduct by people in positions of authority’ that Pichai references in his response seem to vary based on how valuable the perpetrator is to Google or its board. What measures are they taking to ensure an impartial assessment happens? The incidents also highlight that executives brazenly misbehave with their victims. There is no mention of how that aspect of Google is being tackled. Specifically, for example, would Mr. Page take a different decision today if had a chance to go back in time or if Mr. Pichai, as Google CEO personally taken a public stance on specific incidents of sexual misconduct without hiding behind aggregate numbers and figures. The report throws light on the pervasive sexist culture in male-dominated Silicon Valley and the growing chorus denouncing it.  It is traumatic enough to experience such harassment, imagine the pressure that one has to deal with when such incidents go public. It is also sad that the tech giant that everyone looks up to- Google- decided to sweep matters under the carpet to save itself from public attention. These recurring stories seem to have led to the release of Brotopia: Breaking Up The Boys Club of Silicon Valley, a book by Emily Chang, Bloomberg reporter, that dives into the stories of women who say they have been sexually harassed at tech companies and venture capital firms. You can head over to The New York Times for the entire news coverage as well as similar incidents documented. NIPS Foundation decides against name change as poll finds it an unpopular superficial move; instead increases ‘focus on diversity and inclusivity initiatives’ Python founder resigns – Guido van Rossum goes ‘on a permanent vacation from being BDFL’ Ex-googler who quit Google on moral grounds writes to Senate about company’s “Unethical” China censorship plan

Team Tableau Public Kristin Adderson December 23, 2020 - 9:01pm December 24, 2020 Let’s be real, 2020 has been one incredibly wild ride that no one expected. Despite it all, one thing remained steadfast: the Tableau Public global community of data enthusiasts’ commitment to bringing impactful (and often mindblowing) data insights to life. To mark the end of 2020, we’re taking a look back at some of the most amazing visualizations created by the #DataFam this year.  We looked back at highlights from this year’s featured visualizations. Our “Viz of the Day” gallery represents the many ways our community uses Tableau Public to visualize the data topics they’re most passionate about. Each day, the Tableau Public team selects and features a “Viz of the Day” (VOTD) based on a variety of criteria. A viz might tell a clear and compelling story. Perhaps it is visually stunning or includes an innovative chart type. Or, the viz might result from one of the community’s social data projects or competitions. Whatever the reason, each featured viz shares a common trait—demonstrating the realm of possibility when using data to express oneself.  There were over 200 visualizations featured as “Viz of the Day” in 2020. The Tableau Public team reviewed each one and hand-picked our favorite from each month. We’ve strived to highlight a diversity of visualizations with different chart-types on a wide range of topics from authors across the globe. Read about each one, then click on the thumbnail to see each creation in its full glory. See a viz that you love? Don’t forget to let that author know by logging in and “favoriting” it.    JANUARY  The 2019 Global Multidimensional Poverty Index by Lali Jularbal The Multidimensional Poverty Index (MPI) takes an in-depth look at how people experience three dimensions of poverty—Health, Education, and Living Standards—in 101 developing countries. Lali Jularbal visualizes the developing country’s rankings by MPI, Intensity, Headcount, and poverty dimensions. Favorite this viz   FEBRUARY Racial Integration in U.S. Schools by Candra McCrae Desegregation orders were implemented by the Supreme Court to help eliminate segregation in schools across the United States. However, according to a recent Gallup Poll, 57% of U.S. adults believe school segregation is still a moderate or severe problem. In this visualization, Candra McRae looks at the history of racial integration in U.S. schools and explores ideas that could help reduce segregation. Favorite this viz   MARCH Popular Pizza Toppings by Amy Tran Whether or not pineapple belongs on a pizza was arguably one of the most controversial debate topics in 2020. Dig into this #MakeoverMonday visualization by Amy Tran to learn about the most popular pizza toppings in Britain. Did your favorite topping make the list? Favorite this viz   APRIL The World's Dependence on the Travel Industry by Chantilly Jaggernauth The travel and tourism industry accounted for more than 10% of the world’s Gross Domestic Product (GDP) in 2019. Explore this visualization by Chantilly Jaggernauth to see the amount of GDP generated by travel and tourism, including hotels, airlines, travel agencies, and more, in various countries across the globe. Favorite this viz   MAY Teladoc Health, Inc. by Praveen P Jose  Many countries around the world are still struggling to control the spread of coronavirus (COVID-19). As a result, telemedicine has become more popular than ever before. In this visualization, Praveen P Jose looks at the stock price of leading telemedicine provider Teladoc over the last five years. Favorite this viz   JUNE Exonerations in America by JR Copreros Over 2,500 wrongful convictions have been reversed in the U.S. since 1989. Using data from the National Registry of Exonerations, JR Copreros visualizes exonerations by race, state, type of crime, and more, revealing systemic flaws in the criminal justice system. Favorite this viz   JULY Economic Empowerment of Women by Yobanny Samano According to the World Bank, the Women, Business and the Law (WBL) Index, composed of eight indicators, "tracks how the law affects women at various stages in their lives, from the basics of transportation to the challenges of starting a job and getting a pension." In this #MakeoverMonday visualization, Yobanny Samano looks at the WBL index scores for 190 countries. Favorite this viz   AUGUST Constellations Born of Mythology by Satoshi Ganeko How did constellations get their names? Many of them are named after figures in Greek and Roman mythology. Brush up on your stargazing skills and explore this #IronQuest visualization by Satoshi Ganeko to learn about each one. Favorite this viz   SEPTEMBER The Day Lebanon Changed by Soha Elghany and Fred Najjar On August 4, 2020, a large amount of ammonium nitrate stored at the port city of Beirut exploded, killing over 200 people and causing billions of dollars in damage. Soha Elghany and Fred Najjar collaborated to create this visualization, which shows the impact of one of the most powerful non-nuclear explosions in history. Favorite this viz   OCTOBER The Air We Breathe by Christian Felix According to the latest data from the World Health Organization (WHO), 97% of cities in low- and middle-income countries with more than 100,000 inhabitants do not meet WHO air quality guidelines. In this visualization, #IronViz Champion Christian Felix explores the correlation between breathing air inequality and wealth inequality. Favorite this viz   NOVEMBER The Most Popular Dog Breeds by Anjushree B V In 2019, the Pembroke Welsh Corgi made it onto the Top 10 Most Popular Dog Breeds list for the first time. Check out this visualization by Anjushree B V to learn how each dog breed's popularity has changed over time. Favorite this viz   DECEMBER Giant Pandas Overseas by Wendy Shijia  Did you know that China rents out its pandas? Today, over 60 giant pandas, native to south-central China, can be found worldwide. Dive into this visualization by Wendy Shijia to learn when each panda living abroad will be returned to its home country. Favorite this viz   And that’s a wrap! Cheers to an incredible year, made possible by Tableau Public users like you. Be sure to subscribe to “Viz of the Day” to get more visualizations like these—another year’s worth of awe-inspiring community-created data inspiration awaits.  Craving more viz-spiration? Check out these resources commemorating Tableau Public’s 10th anniversary: Ten most-favorited vizzes to celebrate ten viz-tastic years of Tableau Public Ten years later—What Tableau Public means to our community and the world If Data Could Talk: A walk down memory lane with Tableau Public

Sarah Hinrichsen Services Delivery Enablement Manager, Tableau Kristin Adderson December 28, 2020 - 6:13pm December 28, 2020 Okay, so maybe you aren’t completely sold on virtual instructor-led training yet. I am here to reassure you. There’s a place for it in your Tableau education plan. If you haven’t taken virtual training before, the unknown might make you question whether it’s right for you. Or, if you have taken virtual training, you may wonder what Tableau’s has to offer. Let’s ease your concerns and get you excited about attending a Tableau virtual training. Convenient and Flexible Who doesn’t love working from home in their pajamas? Tableau virtual instructor-led training allows you to grab your first cup of coffee (or fifth, we’re not counting) and leisurely get to your computer before class starts. No worrying about getting out the door and commuting—you can join right from home. You can even attend from the beach, lake, coffee shop, or pub down the street. As long as you have internet access, you can join a virtual training class. *Screenshot above is a subset of options. Times are shown in the user's local time. We offer classes in time zones worldwide, so there will be options in or close to your time zone. This allows you the convenience of taking a course during the workday or in a different time zone, so it doesn’t interfere with your daily work. We also offer virtual classes of varying lengths. Some are full days (9 am-5 pm), and some are partial days (2 - 3 hours/day). This gives you the flexibility to complete the course in a couple of days or spread it out over a week. Either way, you are getting the same proven Tableau curriculum delivered by an expert certified instructor. The only downside is that you don’t get a catered lunch. But it also means you can have potato chips and chocolate for lunch—no judgment here. Interactive and Engaging I don’t know about you, but for me, it’s hard to focus when a presenter doesn’t engage with the audience. In a virtual environment, this becomes even more important; no one wants to be talked at for multiple hours in a day. Tableau’s virtual training instructors utilize many different tools to engage with you throughout the class—in the form of a verbal question, a multiple-choice poll, or a hands-on discovery task. You have the opportunity to use the chat or anonymous poll to answer questions. Having multiple outlets for interactivity and switching between the lecture, interactivity, and hands-on activity will allow you to interact more comfortably and keep the class moving.  Not only do our instructors use interactivity to keep you engaged, but they also throw in fun ways to learn the concepts. For example, to make complex topics more relatable—they may use ice cream sales to help you understand Scatter plots or superheroes to discuss Sets and combined Sets or even cookie recipes to get you more familiar with Relationships. Once you understand the concepts at a fundamental level, the instructor relates them to industry-specific examples so you will know how to apply them to your work. In a perfect world, I would want instructors to ask me questions to keep me engaged, and somewhere I can get my questions answered in real-time. Guess what?!  Our virtual training instructors don’t follow a script and answer any questions you might have throughout the course. You can ask questions about the content the instructor is demonstrating. Most importantly, if you ever get stuck during a hands-on activity, the instructor will be able to clarify steps, so that you can move ahead seamlessly. And remember, if you don’t want to come off mute to ask a question, the virtual training environment has a chat feature to chat directly with the instructor. Support Technology doesn’t come without its pitfalls, and many times it can be daunting to have to use new systems and technologies. Our Global Services team is here to help. A week before your training session, you will receive all of the information to download the course materials and log in. You can download the materials ahead of time and test your connection to the virtual environment before the training starts. This not only allows you to stroll into class a few minutes early with peace of mind but if you have any issues before class begins, you can get help from Global Services. They can assist with troubleshooting or even use other resources and technologies to make sure you have the content and information you need.  Not only do you have Global Services to help with technical issues as your first line of defense, but you also have the certified instructor to help. The instructor knows the technologies’ ins and outs and can point you in the right direction if something happens during class. Many of our instructors are seasoned experts and have likely dealt with all the technical issues you throw at them.  Now that you know a little more about virtual instructor led-training at Tableau, you are ready to take your learning experience to the next level! Once you get into the virtual classroom, you will quickly understand how fun and engaging our virtual classes and instructors are. Convinced and ready to sign up? Go to the Live Virtual Training Classes page to see courses and a full schedule. You can register for a class directly from there! Still not convinced? Check out the video at the bottom of the Instructor-Led Training page that gives you a window into our virtual instructor-led training with snippets of live classroom experiences.

On Monday, Intel announced the Pohoiki Beach, a neuromorphic system comprising of 8 million neurons, multiple Nahuku boards and 64 Loihi research chips. The Intel team unveiled this new system at the DARPA Electronics Resurgence Initiative Summit held in Detroit. Intel introduced Loihi in 2017, its first brain inspired neuromorphic research chip. Loihi applies the principles found in biological brains to computer architectures. It enables users to process information up to 1,000 times faster and 10,000 times more efficiently than CPUs for specialized applications like sparse coding, graph search and constraint-satisfaction problems.The Pohoiki Beach is now available for the broader research community and they can experiment with Loihi. “We are impressed with the early results demonstrated as we scale Loihi to create more powerful neuromorphic systems. Pohoiki Beach will now be available to more than 60 ecosystem partners, who will use this specialized system to solve complex, compute-intensive problems,” says Rich Uhlig, managing director of Intel Labs. According to Intel, Pohoiki Beach will enable researchers to efficiently scale novel neural inspired algorithms such as sparse coding, simultaneous localization and mapping (SLAM) and path planning. The Pohoiki Beach system is different in a way because it will demonstrate the benefits of a specialized architecture for emerging applications, including some of the computational problems hardest for the internet of things (IoT) and autonomous devices to support. By using this type of specialized system, as opposed to general-purpose computing technologies, Intel expects to realize orders of magnitude gains in speed and efficiency for a range of real-world applications, from autonomous vehicles to smart homes to cybersecurity. Pohoiki Beach will mark a major milestone in Intel’s neuromorphic research, as it will lay the foundation for Intel Labs to scale the architecture to 100 million neurons later this year. Rich Uhlig says he, “predicts the company will produce a system capable of simulating 100 million neurons by the end of 2019. Researchers will then be able to apply it to a whole new set of applications, such as better control of robot arms.” Ars Technica writes that Loihi, the underlying chip in Pohoiki Beach consists of 130,000 neuron analogs—hardware-wise, this is roughly equivalent to half of the neural capacity of a fruit fly. Pohoiki Beach scales that up to 8 million neurons—about the neural capacity of a zebrafish. But what perhaps is more interesting than the raw computational power of the new neural network is how well it scales. “With the Loihi chip we’ve been able to demonstrate 109 times lower power consumption running a real-time deep learning benchmark compared to a GPU, and 5 times lower power consumption compared to specialized IoT inference hardware. Even better, as we scale the network up by 50 times, Loihi maintains real-time performance results and uses only 30 percent more power, whereas the IoT hardware uses 500 percent more power and is no longer real-time,” says Chris Eliasmith, co-CEO of Applied Brain Research and professor at the University of Waterloo As per the IEEE Spectrum, Intel and its research partners are just beginning to test what massive neural systems like Pohoiki Beach can do, but so far the evidence points to even greater performance and efficiency, says Mike Davies, director of neuromorphic research at Intel. “We’re quickly accumulating results and data that there are definite benefits… mostly in the domain of efficiency. Virtually every one that we benchmark…we find significant gains in this architecture,” he says. Going from a single-Loihi to 64 of them is more of a software issue than a hardware one. “We designed scalability into the Loihi chip from the beginning,” says Davies. “The chip has a hierarchical routing interface…which allows us to scale to up to 16,000 chips. So 64 is just the next step.” According to Davies, Loihi can run networks which are immune to catastrophic forgetting and can learn more like humans. He proved this with an evidence of research work done by the Thomas Cleland’s group at Cornell University, that Loihi can achieve one-shot learning. That is, learning a new feature after being exposed to it only once. Loihi can also run feature-extraction algorithms immune to the kinds of adversarial attacks that can confuse image recognition systems. Traditional neural networks don’t really understand the features they’re extracting from an image in the way our brains do. “They can be fooled with simplistic attacks like changing individual pixels or adding a screen of noise that wouldn’t fool a human in any way,” Davies explains. But the sparse-coding algorithms Loihi can run work more like the human visual system and so wouldn’t fall for such shenanigans. This news brings a lot of excitement amongst the community and they are awaiting to see a system that will contain 100 million neurons by the end of this year. https://twitter.com/javiermendonca/status/1151131213576359937 https://twitter.com/DSakya/status/1150988779143880704 Intel discloses four new vulnerabilities labeled MDS attacks affecting Intel chips Intel plans to exit from the 5G smartphone modem business, following the Apple Qualcomm dispute Google researchers present Zanzibar, a global authorization system, it scales trillions of access control lists and millions of authorization requests per second

On 11th November, the Google Brain Team released Magenta studio in beta, a suite of free music-making tools using their machine learning models. It is a collection of music plugins built on Magenta’s open source tools and models. These tools are available both as standalone Electron applications as well as plugins for Ableton Live. What is Project Magenta? Magenta is a research project which was started by some researchers and engineers from the Google Brain team with significant contributions from many other stakeholders. The project explores the role of machine learning in the process of creating art and music. It primarily involves developing new deep learning and reinforcement learning algorithms to generate songs, images, drawings, and other materials. It also explores the possibility of building smart tools and interfaces to allow artists and musicians to extend their processes using these models. Magenta is powered by TensorFlow and is distributed as an open source Python library. This library allows users to manipulate music and image data which can then be used to train machine learning models. They can generate new content from these models. The project aims to demonstrate that machine learning can be utilized to enable and enhance the creative potential of all people. If the Magenta studio is used via Ableton, the Ableton Live plugin reads and writes clips from Ableton's Session View. If a user chooses to run the studio as a standalone application, the standalone application reads and writes files from a users file system without requiring Ableton. Some of the demos include: #1 Piano Scribe Many of the generative models in Magenta.js requires the input to be a symbolic representation like Musical Instrument Digital Interface (MIDI). But now, Magenta Converts raw audio to MIDI using Onsets and Frames which  a neural network trained for polyphonic piano transcription. This means that only audio is enough to obtain an output of MIDI in the browser. #2 Beat Blender The Beat Bender is built by Google Creative Lab using MusicVAE. Users can now generate two dimensional palettes of drum beats and draw paths through the latent space to create evolving beats. #3 Tenori-of Users can utilize the Magenta.js to generate drum patterns when they hit the “Improvise” button. This is more like a take on an electronic sequencer. #4 NSynth Super This is machine learning algorithm using deep neural network to learn the characteristics of sounds, and then create a completely new sound based on these characteristics. NSynth synthesizes an entirely new sound using the acoustic qualities of the original sounds. For instance, users can get a sound that’s part flute and part sitar all at once. You can head over to the Magenta Blog for more exciting demos. Alternatively, head over to magenta.tensorflow.org to read more about this announcement. Worldwide Outage: YouTube, Facebook, and Google Cloud go down affecting thousands of users Intel Optane DC Persistent Memory available first on Google Cloud Google Cloud Storage Security gets an upgrade with Bucket Lock, Cloud KMS keys and more