Tech News - Data

US Lawmakers have initiated drafting privacy regulations and also encouraging the enforcement agencies to build a privacy framework which they can easily follow. Last week, Marco Rubio, U.S. Senator introduced a bill titled ‘American Data Dissemination (ADD) Act’ for creating federal standards of privacy protection for large companies like Google, Amazon, and Facebook. However, this bill largely focuses on data collection and disclosure. Hence, the experts were afraid that this bill would ignore the way companies use customer’s data. Last week, U.S. Senators John Kennedy and Amy Klobuchar introduced the ‘Social Media Privacy and Consumer Rights Act’ that allows the consumers to have more control over their personal data. This legislation aims to improve transparency, strengthen consumers’ recourse options during a data breach and ensure companies are compliant with privacy policies that protect consumers. Another bill, sponsored by Reps. Dutch Ruppersberger, Jim Himes, Will Hurd, and Mike Conaway, was introduced last week to combat theft of U.S. technologies by state actors including China, and to reduce risks to “critical supply chains.” Ruppersberger said they had long suspected Beijing is using its telecom companies to spy on Americans and they knew that China is responsible for up to $600 billion in a theft of U.S. trade secrets. Some reintroduced bills Securing Energy Infrastructure Act A bill titled ‘Securing Energy Infrastructure Act’ was proposed by Sens. Jim Risch, and Angus King. This bill, reintroduced last Thursday, would push the government to explore new ways to secure the electric grid against cyber attacks. This bill unanimously passed the Senate in December but was never put to a vote in the House. Telephone Robocall Abuse Criminal Enforcement and Deterrence Act On 17th January, Sens. John Thune, R-S.D., and Ed Markey, D-Mass., renewed their call to increase punishments for people running robocall scams. The Telephone Robocall Abuse Criminal Enforcement and Deterrence, or TRACED, Act would give the Federal Communications Commission more legal leeway to pursue and prosecute robocallers. Under the bill, telecom companies would also need to adopt tools to sift out robocalls. Thune said, “The TRACED Act holds those people who participate in robocall scams and intentionally violate telemarketing laws accountable and does more to proactively protect consumers who are potential victims of these bad actors.” Federal CIO Authorization Act The Federal CIO Authorization Act, which Reps. Will Hurd, and Robin Kelly, reintroduced on Jan. 4, passed the House unanimously on Tuesday. This bill would elevate the federal chief information officer within the White House chain of command and designate both the federal CIO and federal chief information security officer as presidentially appointed positions. The measure still lacks a Senate counterpart. Lawmakers have also sent letters to different companies including Verizon, T-Mobile, Sprint, and AT&T asking for information on the companies’ data sharing partnerships with third-party aggregators. These companies have time until Jan 30 to respond. Reps. Greg Walden, Cathy McMorris Rodgers, Robert Latta, and Brett Guthrie, wrote, “We are deeply troubled because it is not the first time we have received reports and information about the sharing of mobile users’ location information involving a number of parties who may have misused personally identifiable information.” To know more about these bills in detail, visit the Nextgov website. Russia opens civil cases against Facebook and Twitter over local data laws Harvard Law School launches its Caselaw Access Project API and bulk data service making almost 6.5 million cases available Senator Ron Wyden’s data privacy law draft can punish tech companies that misuse user data

The TensorFlow team released the first release candidate of TensorFlow 1.13.0-rc0 yesterday. TensorFlow 1.13.0-rc0 explores major bug fixes, improvements and other changes. Let’s have a look at the major highlights in TensorFlow 1.13.0-rc0. Major improvements In TensorFlow 1.13.0-rc0, TensorFlow Lite has been moved from contrib to core. What this means is that Python modules are now under tf.lite and the source code is now under tensorflow/lite instead of tensorflow/contrib/lite. TensorFlow GPU binaries have now been built against CUDA 10. NCCL has been moved to core in TensorFlow 1.13.0-rc0. Behavioural and other changes Conversion of python floating types to uint32/64 (i.e. matching behaviour of other integer types) in tf.constant has been disallowed in TensorFlow 1.13.0-rc0. Doc consisting of details about the rounding mode used in quantize_and_dequantize_v2 has been updated. The performance of GPU cumsum/cumprod has been increased by up to 300x. Support has been added for weight decay in most TPU embedding optimizers such as AdamW and MomentumW. An experimental Java API has been added for injecting TensorFlow Lite delegates. New support is added for strings in TensorFlow Lite Java API. tf.spectral has been merged into tf.signal for TensorFlow 2.0. Bug fixes tensorflow::port::InitMain() now gets called before using the TensorFlow library. Programs that fail to do this are not portable to all platforms. saved_model.loader.load has been deprecated and is replaced by saved_model.load. Saved_model.main_op has also been deprecated and is replaced by saved_model.main_op in V2. tf.QUANTIZED_DTYPES has been deprecated and is changed to tf.dtypes.QUANTIZED_DTYPES. sklearn imports has been updated for deprecated packages. confusion_matrix op is now exported as tf.math.confusion_matrix instead of tf.train.confusion_matrix. An ignore_unknown argument is added in TensorFlow 1.13.0-rc0 to parse_values that suppresses ValueError for unknown hyperparameter types. Such * Add tf.linalg.matvec convenience function. tf.data.Dataset.make_one_shot_iterator() has been deprecated in V1 and added tf.compat.v1.data.make_one_shot_iterator()`. tf.data.Dataset.make_initializable_iterator() is deprecated in V1, removed it from V2, and added another tf.compat.v1.data.make_initializable_iterator(). The XRTCompile op is can now return the ProgramShape resulted from the XLA compilation as a second return argument. XLA HLO graphs are rendered as SVG/HTML in TensorFlow 1.13.0-rc0. For more information, check out the complete TensorFlow 1.13.0-rc0 release notes. TensorFlow 2.0 to be released soon with eager execution, removal of redundant APIs, tf function and more Building your own Snapchat-like AR filter on Android using TensorFlow Lite [ Tutorial ] TensorFlow 1.11.0 releases

Uber and Lyft drivers yesterday went on strike across Los Angeles in opposition to Uber's decision to cut rates by 25% in the Los Angeles area. Organized by Rideshare Drivers United, the strike comes as a further labor-led fightback against ride-hailing platforms following news that U.K.-based Uber drivers are suing the company over access to personal data. If anyone thought 2018's techlash was over, they need to think again - it appears worker solidarity is only getting stronger in the tech industry. What was the purpose of the March 25 Uber strike? Uber and Lyft drivers have experienced declining wages for a number of years as the respective platforms compete to grow customers. This made news earlier in March that Uber would be reducing the per mile rate for drivers from 80¢ to 60¢ particularly tough to take. It underlined to many drivers that things are probably only going to continue to get worse while power rests solely on the side of the platforms. https://twitter.com/_drivers_united/status/1107745851890253824?s=20 But there was more at stake than just an increase in wages. In many ways opposition to Uber's pay cut is simply a first step in a longer road towards improved working conditions and greater power. Rideshare Drivers United actually has an extensive list of aims and demands: A 10% cap on commission The right to organize and negotiate for improved working conditions Ensuring Uber and Lyft are working in accordance with authorities on green initiatives With New York City authorities taking steps to implement minimum pay requirements in December 2018, the action on the west coast could certainly be seen as an attempt to push for consistency across the country. However, it doesn't appear that Los Angeles authorities are interested in taking similar steps at the moment. Uber and Lyft's response to the Los Angeles strike In a statement given to The Huffington Post, an Uber spokesperson said that the changes "will make rates comparable to where they were in September, while giving drivers more control over how they earn by allowing them to build a model that fits their schedule best.” In the same piece, the HuffPo quotes a Lyft spokesperson who points out that the company hasn't changed their rates for 12 months. Support for striking Uber and Lyft drivers Support for the strikers came from many quarters, including the National Union of Health Workers and Senator Bernie Sanders. "One job should be enough to make a decent living in America" the NUHW said. https://twitter.com/NUHW/status/1110270149309849600?s=20 Time for Silicon Valley to rethink There's clearly a long way to go if Rideshare Drivers United are going to achieve their aims. But the conversation is shifting and many Silicon Valley executives will need to look up and take notice - perhaps it's time to rethink things.

Seaborn, the popular data visualization library, has become a very timely and relevant tool for data professionals seeking to enhance their data visualizations. The team behind Seaborn realizes this and hence have pushed the release of Seaborn v0.9.0. This version is a major release with several substantial features and notable API name changes for better consistency with matplotlib 2.0. Three new relational plots Seaborn v0.9.0 features three new plotting functions relplot(), scatterplot(), and lineplot(). These functions bring the high-level API of categorical plotting functions to more general plots. They can visualize a relationship between two numeric variables and map up to three additional variables by modifying hue, size, and style semantics. replot() is a figure-level interface to the two plotting functions and combines them with a FacetGrid. The lineplot() function has support for statistical estimation and is replacing the older tsplot function. It is also better aligned with the API of the rest of the library and more flexible in showing relationships across additional variables. For a detailed explanation of these functions with examples of the various options, go through the API reference and the relational plot tutorial. Notable API name changes Seaborn has renamed a few functions and made changes to their default parameters. The factorplot function has been renamed to catplot(). The catplot() function shows the relationship between a numerical and (one or more) categorical variable using one of several visual representations. This change is expected to make catplot() easy to discover and to define its role better. The lvplot function has been renamed to boxenplot(). The new name makes the plot more discoverable by describing its format (it plots multiple boxes, also known as “boxen”). The size parameter to height is renamed in multi-plot grid objects (FacetGrid, PairGrid, and JointGrid) along with functions that use them (factorplot, lmplot(), pairplot(), and jointplot()). This is done to avoid conflicts with the size parameter that is used in scatterplot and lineplot functions and also makes the meaning of the parameter a bit clearer. The default diagonal plots in pairplot() are changed to now use func:kdeplot` when a "hue" dimension is used. Also, the statistical annotation component of JointGrid is deprecated. Themes and palettes updates Several changes have been made to the seaborn style themes, context scaling, and color palettes to make them more consistent with the style updates in matplotlib 2.0. Here are some of the changes: Some axes style()/plotting context() parameters have been reorganized and updated to take advantage of improvements in the matplotlib 2.0 update. The seaborn palettes (“deep”, “muted”, “colorblind”, etc.) are updated to correspond with the new 10-color matplotlib default. A few individual colors have also been tweaked for better consistency, aesthetics, and accessibility. The base font sizes in plotting context() and scaling factors for "talk" and "poster" contexts have been slightly increased. Calling set() will now call set color codes() to re-assign the single letter color codes by default. Apart from that, the introduction to the library in the documentation has been rewritten to provide more information and critical examples. These are just a select few major updates. For a full list of features, upgrades, and improvements, read the changelog. What is Seaborn and why should you use it for data visualization? Visualizing univariate distribution in Seaborn 8 ways to improve your data visualizations

Amazon re:Invent 2018 commenced yesterday at Las Vegas. This five-day event will comprise of various sessions, chalk talks, and hackathons covering AWS core topics. Amazon is also launching several new products and making some crucial announcements. Adding to this list, yesterday, Amazon announced that AWS Snowball Edge will now come with two options: Snowball Edge Storage Optimized and Snowball Edge Compute Optimized. Snowball Edge Compute Optimized, in addition to more computing power, comes with an optional GPU support. What is AWS Snowball Edge? AWS Snowball Edge is a physical appliance that is used for data migration and edge computing. It supports specific Amazon EC2 instance types and AWS Lambda functions. With Snowball Edge, customers can develop and test in AWS. The applications can then be deployed on remote devices to collect, pre-process, and return the data. Common use cases include data migration, data transport, image collation, IoT sensor stream capture, and machine learning. What is new in Snowball Edge? Snowball Edge will now come in two options: Snowball Edge Storage Optimized: This option provides 100 TB of capacity and 24 vCPUs, well suited for local storage and large-scale data transfer. Snowball Edge Compute Optimized: There are two variations of this option, one is without GPU and the other is with GPU. Both come with 42 TB of S3-compatible storage and 7.68 TB of NVMe SSD storage. You will also be able to run any combination of the instances that consume up to 52 vCPUs and 208 GiB of memory. The main highlight here is the support for an optional GPU. With Snowball Edge with GPU, you can do things like real-time full-motion video analysis and processing, machine learning inferencing, and other highly parallel compute-intensive work. In order to gain access to the GPU, you need to launch an sbe-g instance. You can select the “with GPU” option using the console: Source: Amazon The following are the specifications of the instances: Source: Amazon You can read more about the re:Invent announcements regarding Snowball Edge on AWS website. AWS updates the face detection, analysis and recognition capabilities in Amazon Rekognition AWS announces more flexibility its Certification Exams, drops its exam prerequisites Introducing Automatic Dashboards by Amazon CloudWatch for monitoring all AWS Resources

Activist groups including Human Rights Watch and Amnesty International have spoken out against Apple and Google, for hosting a Saudi Government app, called Absher, that permits the Saudi men to control and decide where the women can travel. As per the complaints of rights groups, Absher promotes discrimination against women and is enforcing ‘gender apartheid’ in Saudi Arabia. This is why they want Apple and Google to consider ‘rehosting’ the app, reports INSIDER. “We call on these companies to assess the risk of human rights abuses and mitigate the harm that these apps may have on women. This is another example of how the Saudi Arabian government has produced tools to limit women's freedoms”, said Dana Ahmed, Saudi Arabia researcher for Amnesty International. Absher app is based on Saudi “guardian” law, according to which, every woman has a legal "guardian" to whom she remains legally dependent for many aspects of life, irrespective of her age, education level or marital status. This male guardian could be her father, uncle, husband, brother, or son, who offers his consent to a variety of basic needs of a woman such as education, clothing, work, money, travel, marriage, etc.                                                                                                                  Absher app (Google Play store) Absher comes with a set of features aimed to restrict women’s travel to specific airports and routes, making sure that in case the woman decides to flee from the country without permission, she can get instantly caught. This is because it comes with an automatic SMS feature that is sent to a woman’s ‘guardian’ for times she crosses borders or makes airport check-ins without permission. 1,000 women try to flee away from Saudi Arabia each year, and text alerts make it very difficult for these women to flee with most of them getting caught by their family members.   The SMS alerts were made compulsory in 2012, however, it received a heavy backlash by the Saudis on social media. This later led to the Saudi Government suspending the SMS alerts in 2014, however, the rights groups believe that the system is still in function. According to Amnesty International, the automated SMS alerts are “another example of how the Saudi Arabian government has produced tools to limit women's freedoms”. Men can also specify the destinations that the women are allowed to travel along with time period for the travel on Absher. Although there are other basic and harmless features in Absher that allows you to pay parking fines, or renew a driver's license, it is mostly used to keep a tight leash on Saudi women.                                                    Absher features What’s even more distressing is the fact that Absher app has been downloaded more than 1 million times on Android devices. Rothna Begum, Middle East researcher for Human Rights Watch told INSIDER, that “Apple and Google have rules against apps that facilitate threats and harassment. Apps like this one can facilitate human rights abuses, including discrimination against women." Apple and Google haven’t responded to the news yet. Public reaction to this news is largely negative with the majority of the people condemning the app and its widespread use in Saudi Arabia: https://twitter.com/Shadow0pz/status/1095030573976961024 https://twitter.com/MustacheofDeath/status/1095186423471210496 https://twitter.com/JuliaFelly/status/1094928811509104642 https://twitter.com/SanctionSaudi/status/1095016257928265729 Read the full story on INSIDER. An AI startup now wants to monitor your kids’ activities to help them grow 'securly' Babysitters now must pass Perdictim’s AI assessment to be “perfect” to get the job Twitter blocks Predictim, an online babysitter-rating service, for violating its user privacy policies; Facebook may soon follow suit

Yesterday, at the ongoing IEEE’s International Solid-State Circuits Conference (ISSCC), Yann LeCun, Facebook AI Research director, presented a paper that touched upon the latest trends and the future of deep learning hardware. ISSCC is a five days event happening in San Francisco, where researchers present the current advances in solid-state circuits and systems-on-a-chip. LeCun in his presentation highlighted several AI trends company should consider in the coming years. Here are some of the highlights from his presentation: Machines should be given some “common sense” With the advancements in deep learning the computer understanding of images, audio, and texts has improved. This has allowed developers to build new applications such as information search and filtering, autonomous driving, real-time language translation, and virtual assistants. These advancements, however, are heavily dependent on supervised learning, which requires human-annotated data or reinforcement learning. LeCun believes that in the next decades, researchers should put their efforts into making machines learn just like humans, by mere observations and occasional actions or in short, by self-supervised manner. To do that, researchers need to find a way to put some level of “common sense” in machines. For this, we need deep learning architectures that are much larger than the one we have currently. LeCun, in his paper Deep Learning Hardware: Past, Present, and Future, wrote, “If self-supervised learning eventually allows machines to learn vast amounts of background knowledge about how the world works through observation, one may hypothesize that some form of machine common sense could emerge.” Empowering machines with human-like capabilities will allow them to make complex decisions. These machines could help in very critical issues like detecting hate speech and inappropriate content on Facebook, enabling virtual assistants to infer context like humans, and more. Ahead of the presentation, LeCun, in an interview with Business Insider said, "There are cases that are very obvious, and AI can be used to filter those out or at least flag for moderators to decide. But there are a large number of cases where something is hate speech but there's no easy way to detect this unless you have a broader context ... For that, the current AI tech is just not there yet." Machine learning chips that can fit everyday devices LeCun is hopeful that in future we will see computer chips that can fit in everyday devices such as vacuum cleaners and lawnmowers. With the machine learning chip incorporated, any device will be able to make smart decisions. For instance, a lawnmower will be able to recognize the difference between weeds and garden roses. Currently, we do have mobile devices with AI built in them to do things like recognizing a user’s face to unlock the device. In the coming years, more work will be put in to make mobile computing chips more sophisticated. LeCun also spoke about the need for hardware specifically designed for deep learning. The current hardware restricts developers to use batches of data in the learning and optimization phase of machine learning models. This will change in the coming years. “If you run a single image, you’re not going to be able to exploit all the computation that’s available to you in a GPU. You’re going to waste resources, basically, so batching forces you to think about certain ways of training neural nets,” he said. A new programming language for deep learning, which is more efficient than Python LeCun believes that deep learning now needs a new programming language which is much more efficient than Python. In an interview with VentureBeat, Yann LeCun said, “There are several projects at Google, Facebook, and other places to kind of design such a compiled language that can be efficient for deep learning, but it’s not clear at all that the community will follow, because people just want to use Python.” He believes that the imaginations of AI researchers and computer scientists tend to be tied to hardware and software tools available. “The kind of hardware that’s available has a big influence on the kind of research that people do, and so the direction of AI in the next decade or so is going to be greatly influenced by what hardware becomes available. It’s very humbling for computer scientists because we like to think in the abstract that we’re not bound by the limitation of our hardware, but in fact, we are.” To know about the other trends LeCun shared, check out the Facebook AI blog. Using deep learning methods to detect malware in Android Applications Researchers introduce a deep learning method that converts mono audio recordings into 3D sounds using video scenes Stanford researchers introduce DeepSolar, a deep learning framework that mapped every solar panel in the US

Last week, in a study published on JAMA Network Open, researchers revealed that machine learning algorithms trained with physical activity data collected from health tracking devices can be used to re-identify actual people. This study indicates that the current practices for anonymizing health information are not sufficient enough. Personal health and fitness data collected and stored by fitness wearable devices can be potentially sold to third parties, like employers, insurance providers, and other companies, without the users’ knowledge or consent. Also, health app makers might be able to link users name to their medical record and then sell this information to third-parties. Location information from activity trackers could be used to reveal sensitive military sites. Therefore, there is a need for a deidentification algorithm that aggregates the physical activity data of multiple individuals to ensure privacy for single individuals. For this study, the researchers analyzed the National Health and Nutrition Examination Survey (NHANES) 2003-2004 and 2005-2006 datasets. These datasets included recordings from physical activity monitors, during both a training run and an actual study mode, for 4,720 adults and 2,427 children. How does the reidentification procedure work? The machine learning model was constructed by building a separate multiclass classifier for each combination of demographic attributes. They used two different machine learning algorithms for multiclass classification, namely, linear support vector machine and random forests. The models were then tested by feeding in the demographic and physical activity data, but not the record numbers, from the testing data into the models to make predictions of record numbers. The accuracy of the models was calculated by counting how many predicted record numbers matched the actual record numbers in the testing data. The following block diagram depicts the steps of this procedure: Source: JAMA Network Open Results of this study The random forest algorithm was able to reidentify the demographic and physical activity data of 4478 adults (94.9%) and 2120 children (87.4%) in NHANES 2003-2004 and 4470 adults (93.8%) and 2172 children (85.5%) in NHANES 2005-2006. The linear SVM algorithm was able to reidentify the demographic and physical activity data of 4043 adults (85.6%) and 1695 children (69.8%) in NHANES 2003-2004 and 4041 adults (84.8%) and 1705 children (67.2%) in NHANES 2005-2006. How privacy risks can be reduced? Per the research paper, the privacy risks posed on individuals by sharing physical data can be reduced by sharing data not only in time but also across individuals of largely different demographics. This is particularly important for governmental organizations such as NHANES that publicly release large national health datasets. Also, currently we do not have strict regulations for organizations that collect and share these sensitive health data. Policymakers should develop regulations to minimize the sharing of activity by device manufacturers. You can go through the research paper for more details: Feasibility of Reidentifying Individuals in Large National Physical Activity Data Sets From Which Protected Health Information Has Been Removed With Use of Machine Learning. Researchers unveil a new algorithm that allows analyzing high-dimensional data sets more effectively, at NeurIPS conference Researchers develop new brain-computer interface that lets paralyzed patients use tablets Facebook AI researchers investigate how AI agents can develop their own conceptual shared language

Microsoft kicked off its highly anticipated Microsoft Connect(); 2017 annual conference on the 14th of November. This three-day annual developer conference is targeted at improving the overall developer experience for building future-oriented apps. In the words of Mitra Azizirad, Corporate VP of Microsoft’s Cloud+Enterprise : “Whether you are creating cloud native-applications, targeting the edge of devices and Internet of Things, infusing your apps with AI, or just getting started, Connect(); 2017 will equip you with the tools and skills you need to build the apps of the future” Key highlights from the Microsoft Connect(); 2017 This year the conference is all about Microsoft forming new partnerships, creating better platforms, enhancing developer productivity, and developing AI enabled tools. A large number of announcements were made pertaining to these areas. New platforms and partnerships Microsoft announced new platforms and partnerships catering to their customers as well as the open source community. Microsoft Azure + DataBricks + Apache Spark = Azure Databricks Microsoft has partnered with Databricks to bring the unique benefits of Apache Spark analytics platform with Databricks in the enterprise cloud.  Termed as Azure Databricks, this analytics platform is optimized for Azure to help data scientists, data engineers, and business decision-makers with streamlined workflows and an interactive workspace. Microsoft joins MariaDB Foundation Microsoft has collaborated with MariaDB community to work closely with the MariaDB foundation. In addition to this, they have also launched a preview of Azure Database for MariaDB service. Developers using Azure Database for MariaDB can now build intelligent apps; Azure Database for PostgreSQL and MySQL already exist. Azure Cosmos DB with Apache® Cassandra API Microsoft has also launched native support for Apache Cassandra API in Azure Cosmos DB. This comes as an integration of Azure Cosmos’s multimodal database service with Cassandra SDKs and tools, without any app code changes. This means developers can now use Cassandra-as-a-service powered by Azure Cosmos DB. GitHub Partnership on GVFS Microsoft has also partnered with Github to manage their large-scale source code repositories. This is made possible through their Git Virtual File System (GVFS) project. Microsoft has built GVFS as an open-source extension to the Git version control system, making it easy to manage over 25 million user repositories. Productivity enhancement As with every year, a key focus area has been to enhance developer productivity, at an individual as well as at a team level. For this the following announcements were made: Azure DevOps Projects Microsoft announced their Azure DevOps project. This will allow developers to build an Azure application on any Azure service using a wide variety of tech stacks. It can also configure a full DevOps pipeline fueled by Visual Studio Team Services. Visual Studio App Center Microsoft has also announced the general availability of its Visual Studio App Center. This app development lifecycle solution helps developers automate, test, manage, distribute, and monitor the lifecycle of their iOS, Android, Windows and macOS apps in the cloud. Visual Studio Live Share Microsoft also unveiled a real-time collaboration tool for developer productivity enhancement. Termed as Visual Studio Live Share, it allows developers using Visual Studio or Visual Studio Code to collaboratively edit and debug their code in real time. It also allows sharing their projects with other developers. Visual Studio Connected Environment for Azure Container Service (AKS) Developers can now use a new connected environment on Microsoft. This would be offered by Azure Container Services(AKS). It would allow developers to easily edit and debug cloud-native applications working on Kubernetes. New Artificial Intelligence tools Artificial Intelligence is revolutionizing how humans interact with technology. With this in mind, Microsoft has announced new AI tools to bring machine learning and intelligence to its developer audience. Azure IoT Edge Microsoft has made available the preview of Azure IoT Edge, a service for building AI applications for the Edge.  Support for AI Toolkit for Azure IoT Edge, Azure Machine Learning, Azure Stream Analytics and Azure Functions is also provided.  Developers can easily build AI applications using Azure Machine Learning and then deploy and manage them on the Azure IoT Edge. Visual Studio Tools for AI Visual Studio Tools for AI is an extension of their Visual Studio IDE. It will allow developers to create, debug, and edit AI applications and scale them to the cloud. It also supports popular deep learning frameworks including Cognitive Toolkit (CNTK), TensorFlow, and Caffe. Key takeaways from the Microsoft Connect(); 2017 This is what we understand from the Microsoft Connect(); 2017 announcements: Microsoft sees partnerships as the key to success, and have partnered with prominent organizations and popular open source communities to help develop better products for their consumers and improve the overall developer experience by providing them easy to use tools and services. The AI wave is a next big hit for Microsoft, as the top players in the tech world (read Google, IBM, Amazon) have already adopted AI as the weapon of choice. Microsoft is catching up real fast, with the launch of their Visual Studio platform specific to AI application. This is a good move to stand head-to-head among the leaders. Edge computing is the next cutting-edge for Microsoft, as portrayed by their Azure IoT edge service. Bringing the developer community closer, by focusing on providing developers easier ways to collaborate and share their projects. The launch of their real-time collaboration tool, Visual Studio Live Share, and a new Connected environment are the next steps towards this goal. Further announcements are expected in the upcoming days. You can visit our website for further updates on upcoming announcements and detailed analysis. For live coverage, you can tune into Connect(); 2017 for more interesting stuff from Microsoft.

The strange mixture of panic and excitement around artificial intelligence only appears to grow as the world uncovers new and more novel ways of using it. These waves of innovation then only feed into continuing cycles of stories that have a habit of perpetuating misleading ideas about both the threats and opportunities it presents. It shouldn't be surprising, then, that there's a serious misunderstanding of what artificial intelligence really is and how it works - as Rowel Atienza told us last month, "we're still very far from robots taking over society." However, artist Holly Herndon (who, incidentally, is a researcher at Stanford) is getting listeners to think differently about artificial intelligence. On her latest album PROTO, which was released today, she's using it to augment and complement her music. Holly Herndon's AI agent, Spawn The special guest that makes PROTO remarkable is Spawn, an AI agent created by Herndon, her husband, and a software engineer. What makes Spawn particularly significant is that Herndon doesn't use it to replace or recreate something but instead as something that exists alongside human activity and creativity. How does Spawn work? Spawn was 'trained' on the music that Herndon and her band were writing for the album. In essence, then, this makes it quite different from the way in which AI is typically used, in that it was developed around a new dataset, not an existing one. When we use existing data sets - and especially when we use them uncritically, without any awareness of how they reproduce or hide certain biases - the AI develops around those very biases. However, when learning from the new 'data', which bears all the marks of Herndon's creative decision making, Spawn almost becomes a 'creative' AI agent. If you listen to the album, it's not always that easy to spot which parts are created by the artificial intelligence and which are made by human musicians. This combination of creative 'sources' means Herndon's album forces us to ask questions about how we use AI and how it interacts with our lives. It quite deliberately engages with the conversation around ethics in AI that has been taking place across the tech industry over the last year or so. https://open.spotify.com/album/3PkYFFSJTPxOhnSYBtyZsk?si=OgFCY5p4Tu2u2rK-3mFYjA "The advent of sampling raised many questions about the ethical use of material created by others," Herndon wrote in a statement published on Twitter at the end of 2018, "but the era of machine legible culture accelerates and abstracts that conversation." https://twitter.com/hollyherndon/status/1069978436851113985 What does Holly Herndon's album tell us about artificial intelligence? PROTO raises a number of really important questions about artificial intelligence. First and foremost, it suggests that artificial intelligence isn't a replacement for human intelligence. Spawn isn't used to take the jobs from any musicians, but rather extends what's sonically possible. It adds to their capabilities, giving it a new dimension. Furthermore, just as Herndon refuses to see artificial intelligence as something which can replicate human labor - or creativity - it also points out some of the very problems with this kind of understanding: the idea that AI can 'replicate' human intelligence at all. Instead, the album's merging of the human and the artificial is a way of exploring the weaknesses of artificial intelligence. This is a way of making AI more transparent. It opens up something that so seems seamless, and highlights the ways it doesn't quite work. It almost refracts rather than mimics the sound the human musicians make. As Herndon said in an interview with Jezebel publication The Muse, "the technology is impressive and it’s cool but it’s really early still. We really wanted to be honest about that and show its mistakes and show how kind of rough the technology is still because... it's more honest and more interesting, to allow it to have its own aesthetic." https://www.youtube.com/watch?v=r4sROgbaeOs Read next: Why an algorithm will never win a Pulitzer The human side of AI technology But the album does more than just present AI as a flawed tool that can complement human ingenuity. It also asks us about ownership and creativity. It uses the technology as a way of tackling human questions like "what does it mean to create something?" and "who's even allowed to create things?" This is important when we consider the fact that not only does someone control and own a given algorithm - as in literally owning the intellectual property - but also that someone owns and controls the swathes of data that are, at a really fundamental level, crucial to artificial intelligence being possible at all. "The history of music and our shared, human, intellectual project that leads up to today, is a shared resource that we all tap into and we all learn from," Herndon also said in the interview with Jezebel. "So if an individual can just scrape that and then claim so much of that as their own because they hold the keys to this AI, and then they can recreate it, of course it’s going to give people anxiety because there’s an ethical issue with that." Read next: Sex robots, artificial intelligence, and ethics: How desire shapes and is shaped by algorithms Instrumental and aesthetic artificial intelligence One of the main reasons artificial intelligence has become a buzzword is because it's a tool for industry. It has a commercial value; it can improve efficiency by allowing us to do more with less. The value of an album like PROTO - even if it's not the sort of thing you'd usually listen to - is that it removes artificial intelligence from a context in which it is instrumentalized, and puts it into one that's purely aesthetic. To make that clearer, it changes something we'd typically think about in a functional manner - is it working? is it doing what it's supposed to do? - to something in which it's very function is open to question. If Herndon's album is able to do that in even the smallest way, then that can only be a good thing, right? And even if it doesn't - at least it sounds good...

In January, San Francisco legislation proposed a ban on using facial recognition technology by the government. The ban is imposed on government agencies, including the city police and county sheriff’s department, but excludes the technology that unlocks the iPhone or cameras installed by businesses or individuals. Again this month, San Francisco Supervisor Aaron Peskin introduced the Stop Secret Surveillance Ordinance. And yesterday it has been reported that the Board of Supervisors voted in favor of the ban to use facial recognition by city agencies. https://twitter.com/UberFacts/status/1128454197324800000 https://twitter.com/SarahNEmerson/status/1128424297003868160 Northern California’s Matt Cagle and Brian Hofer, chair of Oakland’s Privacy Advisory Commission, came in support of this ordinance and wrote in an op-ed last week, “If unleashed, face surveillance would suppress civic engagement, compound discriminatory policing, and fundamentally change how we exist in public spaces.” https://twitter.com/Matt_Cagle/status/1128418575159418880 The proposal faced opposition from few, a local group named Stop Crime SF argued a ban might not be that fruitful when talking about property crime and might also impact in collecting and presenting evidence of crime. Though the vice president of Stop Crime SF, Joel Engardio, seems to be satisfied with the amended bill. In a statement to Wired, he says, “We agree with the concerns that people have about facial ID technology. The technology is bad and needs a lot of improvement.” This move definitely would impact the use of technology all over the world and might motivate other cities to adopt the same. Last month, the Oakland Privacy Advisory Commission released 2 key documents, an initiative to protect Oaklanders’ privacy namely, Proposed ban on facial recognition and City of Oakland Privacy Principles. Techies and developers of the facial recognition systems have been showing their concern in this regard and think that introducing strict rules and surveillance would be better than putting up a ban. Benji Hutchinson, vice president of federal operations for NEC, a major supplier of facial-recognition technology, says, “I think there’s a little bit too much fear and loathing in the land around facial-recognition technology.” In a statement to Wired, Daniel Castro, vice president of the Information Technology and Innovation Forum believes in calling for safeguards on the use of the technology rather than prohibitions. He also calls ban a “step backward for privacy,” as it will leave more people reviewing surveillance video. Though in the board meeting, Peskin said, “I want to be clear — this is not an anti-technology policy.”  He further clarified that the ordinance is also an accountability measure which would ensure safe and responsible use for surveillance tech. Update from ACLU on 21st May San Francisco's ban on using facial recognition technology by the government is now official. Yesterday, Matt Cagle tweeted that San Francisco has approved the ban by a vote of 10 to 1. https://twitter.com/Matt_Cagle/status/1130947088605298688 Amazon finally agrees to let shareholders vote on selling facial recognition software Oakland Privacy Advisory Commission lay out privacy principles for Oaklanders and propose ban on facial recognition China is using facial recognition tech to profile 11 million Uighurs Muslim minority: NYT report  

Over hundred conservative Facebook employees have formed an online group to protest against the company’s “intolerant” liberal culture. First reported by  the New York Times, these employees have formed an internal online group “FB’ers for Political Diversity”,  which is a space for ideological diversity within the company. Brian Amerige, a senior Facebook engineer wrote in the group, “We are a political monoculture that’s intolerant of different views. We claim to welcome all perspectives, but are quick to attack — often in mobs — anyone who presents a view that appears to be in opposition to left-leaning ideology.” Said to follow the principles of Ayn Rand, Mr. Amerige, started working at Facebook in 2012. He posted a 527-word memo about political diversity at Facebook on his personal website on 20 Aug 2018. He also proposed that Facebook employees debate their political ideas in the new group to better equip the company to host a variety of viewpoints on its platform. This activity comes as quite a surprise in Facebook’s largely liberal workplace culture - a rare sign of an organized disagreement. The last few years, Facebook witnessed many disturbing events, from the spread of misinformation by Russians, the mishandling of users’ data, and the ban of Alex Jones. Critics from the group consider these moves to be a sign that Facebook harbors an anti-conservative bias. This new group has received both praise and backlash from Facebook employees. Some say that its online posts were offensive to minorities. Per the New York Times, one engineer (name undisclosed), said “several people had lodged complaints with their managers about FB’ers for Political Diversity and were told that it had not broken any company rules.” However, some Facebook employees considered the group to be constructive and inclusive of different political viewpoints. Facebook is yet to comment on their employees’ political ideology. With Sheryl Sandberg, Facebook’s chief operating officer, scheduled to testify at a Senate hearing about social media manipulation in elections, this protest adds a one more dimension to the social complexities that Facebook often finds itself these days. For more details, read the original post on the New York Times. Facebook’s AI algorithm finds 20 Myanmar Military Officials guilty. Facebook bans another quiz app and suspends 400 more due to concerns of data misuse. Facebook is reportedly rating users on how trustworthy they are at flagging fake news.

Meredith Whittaker who played a major role in Google’s Walkout last year is leaving the company amid facing retaliation at work. The news was disclosed when a software engineer at Google posted a tweet about her last day. https://twitter.com/thegreenfrog611/status/1150859347766833152   A Google spokeswoman also confirmed Whittaker’s departure to Bloomberg. However, Whittaker has not yet shared the news on her Twitter account. Last year in November, a global Google Walkout for Real Change was organized by Claire Stapleton, Meredith Whittaker and six other employees at the company. It prompted 20,000 Google employees and contractors to walk off the job opposing the company’s handling of sexual harassment allegations. In April, Stapleton and Whittaker accused the company of retaliation against them over last year’s Google Walkout protest. Both their roles changed dramatically including calls to abandon AI ethics work, demotion, and more. After the announcement of Google disbanding it’s AI Ethics council, Whittaker said, she was informed that to remain at the company she will have to abandon her work on AI ethics and the AI Now Institute. She said that her manager told her in late December she would likely need to leave Google’s Cloud division. The same manager told her in March that the “Cloud division was seeking more revenue and that AI Now and her AI ethics work was no longer a fit. This was a strange request because the Cloud unit has a team working on ethical concerns related to AI.” Similar retaliation was faced by Stapleton, who was told she would be demoted from her role as marketing manager at YouTube. “My manager started ignoring me, my work was given to other people, and I was told to go on medical leave, even though I’m not sick,” Following continuous counter-attacks, Stapelton was prompted to resign from her position last month. https://twitter.com/clairewaves/status/1137002800053985280 Whittaker had then tweeted in her support. https://twitter.com/mer__edith/status/1137006840313548801 Whittaker had signed the petition protesting Google’s infamous Project Dragonfly, the secretive search engine that Google is allegedly developing which will comply with the Chinese rules of censorship. Meredith Whittaker was also a leader in the anti-Maven movement. Google’s Project Maven, was focused on analyzing drone footage and could have been eventually used to improve drone strikes on the battlefield. More than 3,000 Google employees signed a petition against this project that led to Google deciding not to renew its contract with the U.S. Department of Defense in 2019. Google announced in June it would not renew the contract. Whittaker tweeted at the time that she was “incredibly happy about this decision, and have a deep respect for the many people who worked and risked to make it happen. Google should not be in the business of war.” People have commented on how Meredith's departure will only intensify activism at Google. “The impact @mer__edith has in AI ethics is second to none. What happens to her at Google will be a gauge for the wellbeing of the entire field. Watch closely,” Moritz Hardt, Assistant Professor of Electrical Engineering and Computer Science, Berkeley University https://twitter.com/mrtz/status/1121110692843507712 https://twitter.com/Kantrowitz/status/1150992543691108352 Liz Fong-Jones, Xoogler, who left Google over ethical concerns earlier this year, tweeted about the number of Google Walkout and other organizing leaders that have left the company. There are five who have left, Claire Stapleton, Meredith Whittaker, Liz Fong-Jones, Celie O'Neil-Hart, and Erica Anderson. https://twitter.com/lizthegrey/status/1150960547803860993 Google rejects all 13 shareholder proposals at its annual meeting, despite protesting workers Google Walkout organizer, Claire Stapleton resigns after facing retaliation from management #NotOkGoogle: Employee-led town hall reveals hundreds of stories of retaliation at Google

Comet.ml, is one of the infrastructure-agnostic machine learning (ML) platforms which is simple, fast and free for open source projects. It launched the first platform for data science and machine learning users to track, monitor and optimize their machine learning models. Comet allows data science teams to track their code, experiments, and results on machine learning projects. The newly launched platform allows users to optimize their machine learning and artificial intelligence models and twist hyperparameters of their demonstrations. The platform also provides dashboards which help in collaboration of codes of the ML research and results. It allows researchers to view results with an intuitive graph and compare various aspects and versions of the machine learning experiments. Comet also functions on popular Machine Learning libraries such as Keras, TensorFlow, PyTorch, scikit-learn, and Theano. The platform allows teammates to collaborate real-time without affecting the mobility and adaptability of the datasets and production models. Key Features of Comet: Single-line Tracking - Start tracking with just a single line into your training code. It works on any machine and with any type of model. Compare Experiments - Compare different experiments and observe the code differences, hyper-parameters, and various other data points. Integration with Git - Comet allows to integrate with Github and other git service providers. After finalizing  the experiment, it automatically generates a pull request with the model with the best accuracy to the Github repository. Collaboration - Share multiple projects with team members and stakeholders along with visibility and insights into project team performance. Documentation -  Provides Notes section allowing you to add and manage documentation for all projects and training experiments. Comet is already adopted by more than 30 industry leaders and research universities with more than 6000 large-scale machine learning models. Check out the video to know more about the platform functionality: https://www.youtube.com/watch?v=LlsRMQjV__c&feature=youtu.be Other latest news for a quick read: Deeplearning4j 1.0.0-alpha arrives! How greedy algorithms work?  

Microsoft has released to Insiders the first test construct of its next Windows Server. Build 17035 of Windows Server is identical to its PC counterpart, codenamed "Redstone 4." Assuming Microsoft sticks to its six-month release cadence, both the client and server function updates are anticipated to be designated 1803 (for March 2018), and to start rolling out to mainstream users around April 2018. In Microsoft dictionary, Windows Server 1803 could, therefore, be the next "Semi-Annual Channel release" for Windows Server. As of this construct, Server Insiders get the selection of ISO layout or VHDX layout, with photographs pre-keyed, eliminating the need to input a key right through the setup. An important announcement in 17035 is the return of Storage Spaces Direct (S2D), which mysteriously disappeared from Windows Server 1709, allegedly because of quality issues. Since then, Microsoft has been promising “hyper-converged innovation” in “another release available very soon.” But now the tech leader has said that the software-defined storage tool is coming back with “some new and necessary updates” added to it including Data Deduplication and Resilient File System (ReFS). The only other addition to this construct is that builders can use localhost or loopback (127.0.0.1) to get admission to products and services running in containers on the host. With this new test build, Microsoft is making available Honolulu Technical Preview 1711 Construct 01003. Honolulu is a graphical control device for Windows Server. Inside Honolulu, Microsoft is making updates and adjustments to Remote Desktop, Windows 10 client management, Switch Embedded Teaming, and Data grid performance. In its announcement, Microsoft has listed the recognized problems with Server 17035, such as the cases when the base filtering engine (BFE) service may fail to start preventing the Windows Defender firewall (MpsSvc service) from starting, and that testing of the Windows core may fail because of a timeout while attempting to load the test libraries. For Honolulu Tech Preview 1711, all the known issues have been listed here. The return of Storage Spaces Direct is a definite takeaway. But is this build preview all that “big hyper-converged innovation” Microsoft teased us with? ReFS has been around since Windows Server 2012, and Data Deduplication is a checkbox feature for any storage device. We just have a feeling there could be bigger announcements in store.