Tech News

On Monday, .NET engineering team announced the latest monthly release of their cross-platform, open source machine learning framework for .NET developers, ML.NET 0.6. Some of the exciting features in this release include new API for building and using machine learning models, performance improvements, and much more. Improvements in the ML.NET 0.6 A new LearningPipeline API for building ML model The new API is more flexible and enables new tasks and code workflow that weren’t possible with the previous LearningPipeline API. The team further plans to deprecate the current LearningPipeline API. This new API is designed to support a wider set of scenarios. It closely follows ML principles and naming from other popular ML related frameworks like Apache Spark and Scikit-Learn. Know more about the new ML.NET API, visit the Microsoft blog. Ability to get predictions from pre-trained ONNX Models ONNX, an open and interoperable model format enables using models trained in one framework (such as scikit-learn, TensorFlow, xgboost, and so on) and use them in another (ML.NET). ML.NET 0.6 includes support for getting predictions from ONNX models. This is done by using a new transformer and runtime for scoring ONNX models. There are a large variety of ONNX models created and trained in multiple frameworks that can export models to ONNX format. Those models can be used for tasks like image classification, emotion recognition, and object detection. The ONNX transformer in ML.NET provides some data to an existing ONNX model and gets the score (prediction) from it. Performance improvements In the ML.NET 0.6 release, there are made several performance improvements in making single predictions from a trained model. Two improvements include: Moving the legacy LearningPipeline API to the new Estimators API. Optimizing the performance of PredictionFunction in the new API. Following are some comparisons of the LearningPipeline with the improved PredictionFunction in the new Estimators API: Predictions on Iris data: 3,272x speedup (29x speedup with the Estimators API, with a further 112x speedup with improvements to PredictionFunction). Predictions on Sentiment data: 198x speedup (22.8x speedup with the Estimators API, with a further 8.68x speedup with improvements to PredictionFunction). This model contains a text featurizer, so it is not surprising to see a smaller gain. Predictions on Breast Cancer data: 6,541x speedup (59.7x speedup with the Estimators API, with a further 109x speedup with improvements to PredictionFunction). Improvements in Type system In this ML.NET version, the Dv type system has been replaced with .NET’s standard type system. This makes ML.NET easy to use. ML.NET previously had its own type system, which helped it deal with missing values (a common case in ML). This type system required users to work with types like DvText, DvBool, DvInt4, etc. One effect of this change is, only floats and doubles have missing values which are represented by NaN. Due to the improved approach to dependency injection, users can also deploy ML.NET in additional scenarios using .NET app models such as Azure Functions easily without convoluted workarounds. To know more about other improvements in the ML.NET 0.6 visit the Microsoft Blog. Microsoft open sources Infer.NET, it’s popular model-based machine learning framework Neural Network Intelligence: Microsoft’s open source automated machine learning toolkit .NET Core 3.0 and .NET Framework 4.8 more details announced

Food supply chains have become a contentious topic in recent years, with question marks over standards, safety and the environmental impact of modern farming and manufacturing. IBM, however, would seem to have found a solution - blockchain. The tech multinational yesterday (October 8) launched its Food Trust system, a ledger that will give stakeholders at every point in the supply chain improved transparency over where food has come from and how it has been produced. IBM has been working on its Food Trust system for 18 months. With its launch, it has revealed a number of organizations taking part in the pilot scheme, such as French food giant Carrefour and Walmart. Laurent Vallée, Carrefour's General Secretary, said: "Being a founding member of the IBM Food Trust platform is a great opportunity for Carrefour to strongly accelerate and widen the integration of blockchain technology to our products in order to provide our clients with safe and undoubted traceability." Why blockchain could be a game-changer for the food industry The advantages of using IBM's blockchain-backed technology is clear - it offers transparency for everyone. Because a ledger offers a single view for all stakeholders, every transaction that takes place is observable - this means that bad practices, such as a vendor going to a different supplier from the one agreed, to, say, cut costs, can be easily identified. Walmart's Frank Yiannas, VP for food safety, explained that Walmart wants to "create the equivalence of FedEx tracking for food," so everyone involved in a supply chain has full visibility. Obviously, from a safety perspective, this could be huge, and should, in time at least, discourage bad practice. The success of IBM's Food Trust system remains to be seen. It is, however, a sharp riposte to blockchain sceptics. It might also turn out to be an important victory for tech optimists. At a time when technology appears hell bent on eroding trust, perhaps this is a vital example of technology helping to reassert the importance of trust in the public realm. Read next: How far will Facebook go to fix what it broke: Democracy, Trust, Reality

Microsoft announced, yesterday, that it is collaborating with Grab, the leading on-demand transportation, mobile payments and online-to-offline services platform in Southeast Asia, as part of a strategic cloud partnership. The partnership aims to transform the delivery of digital services and mobility by using Microsoft’s state-of-the-art expertise in machine learning and other artificial intelligence (AI) capabilities. “Our partnership with Grab opens up new opportunities to innovate in both a rapidly evolving industry and growth region. We’re excited to team up to transform the customer experience as well as enhance the delivery of digital services for the millions of users who rely on Grab for safe and affordable transport, food and package delivery, mobile payments, and financial services”, mentioned Peggy Johnson, executive vice president at Microsoft. Grab is a Singapore-based technology company delivering ride-hailing, ride sharing, and logistics services via its app in Singapore and neighboring Southeast Asian nations. It currently operates in 235 cities across eight Southeast Asian countries.  Moreover, Grab’s digital wallet, GrabPay, is the top player in Southeast Asia. This partnership is expected to help both companies explore a wide range of innovative deep technology projects such as mobile facial recognition with built-in AI for drivers and customers, using Microsoft Azure’s fraud detection services to prevent fraudulent transactions on Grab’s platform, and so on. These projects aim to transform the experience for Grab’s users, driver-partners, merchants as well as agents. Grab will be adopting Microsoft Azure as its preferred cloud platform and Microsoft is set to make a strategic investment in Grab; the magnitude of which currently undisclosed. “As a global technology leader, Microsoft’s investment into Grab highlights our position as the leading homegrown technology player in the region. We look forward to collaborating with Microsoft in the pursuit of enhancing on-demand transportation and seamless online-to-offline experiences for users”, said Ming Maa, president of Grab. There are a few other areas of collaborations between Grab and Microsoft. These include Microsoft Outlook integration, Microsoft Kaizala, In-car solutions, and integration of Microsoft Rewards Gift Cards. For more information, check out the official Microsoft blog. Microsoft open sources Infer.NET, it’s popular model-based machine learning framework Microsoft announces Project xCloud, a new Xbox game streaming service, on the heels of Google’s Stream news last week Microsoft’s new neural text-to-speech service lets machines speak like people

Mozilla spoke out about using the word “meritocracy”, as a way to describe its governance and leadership structures, last week. Mozilla has decided to discontinue using the word “meritocracy” following the revised proposal last week. “We have been thinking about the words we use as important carriers of our intended culture and the culture we wish to see in the broader movements we participate in. We have, therefore, taken up a review of our language and practices”, writes Emma Irwin on the Mozilla diversity blog. The first line of Mozilla’s governance reads, “Mozilla is an open source project governed as a meritocracy.” Meritocracy refers to a structure governed by people selected according to merit. On the surface, this sounds ideal, but given its abused usage, it had become analogous to an insidious way of maintaining exclusivity to those with access and experience. Using the term “meritocracy” to refer to communities suffering from a “lack of diverse representation” and without “equal opportunity” seems biased thereby perpetuating a dominant monoculture. Emma states that about 20 years ago, “meritocracy” was the best practice among open source projects. However, now this concept has become more linked to hidden bias and outright abuse. This bias involves devaluing a person’s contributions based on certain aspects of their identity such as gender, race, etc. Many advocates for women and minorities in the tech world have spoken out about how the tech industry considers men to be better at their jobs than everyone else. They don’t take into account the fact that many men have had more opportunities to succeed than their minority and female counterparts. A general belief in the open source community is that those contributing the most, have the most merit and are most deserving. Those who contribute less are not as “meritorious”, ignoring the fact that they might have less access to opportunity, time, and money, preventing them from contributing freely. This is the reality of the open source “meritocracies”, and one of the driving factors behind less diverse culture among the open source community. As a result, Patrick Finch, former strategist, Mozilla, and Emma Irwin, Open Project & Communities Specialist, Mozilla, revised the proposal to better articulate the major principle behind Mozilla’s governance statement. The revised proposal states: Mozilla is an open source project. Our community is structured as a virtual organization. Authority is primarily distributed to both volunteer and employed community members as they show their ability through contributions to the project. The project also seeks to debias this system of distributing authority through active interventions that engage and encourage participation from diverse communities. Mitchell Baker, Mozilla’s co-founder, and chair accepted the revised proposal. She said: “The original meaning I took for meritocracy in open source meant empowering individuals, rather than managers, or manager’s managers or tenure-based authority. However, it’s now clear that so-called meritocracies have included effective forms of discrimination. I personally long for a word that conveys a person’s ability to demonstrate competence and expertise and commitment separate from the job title, or college degree, or management hierarchy, and to be evaluated fairly by one’s peers”. Mozilla is not the only one who has decided to step down from using meritocracy. GitHub’s CEO Chris Wanstrath literally removed the rug at Github’s San Francisco headquarters, back in 2014. The rug kept as a centerpiece at the office room was emblazoned with the phrase, “United Meritocracy of GitHub”. Another example of organizations taking a step to promote diversity is Python dumping the offensive ‘master’, ‘slave’ terms in its documentation. Also, in September, Linux dropped its meritocracy based Code of Conflict in favor of a new Code of Conduct just as Linus Torvalds took a break from Linux to work on his behavioral issues. This new Code of Conduct has been at the epicenter of a major culture clash within the Linux community of late. “I long for a word that makes it clear that each individual who shares our mission is welcome, and valued, and will get a fair deal at Mozilla – that they will be recognized and celebrated for their contributions without regard to other factors. Sadly, “meritocracy” is not the word it once was. The challenge is not to retain a word that has become tainted. The challenge is to build teams and culture and systems that are truly inclusive”, says Baker. People’s opinions about Mozilla’s decision are varied: “Well, meritocracy means we will try to select and give influence/recognition based on merit. When someone tries to push an idea like "let's have more Asians" or "let's have more men" in the project we can direct them to the very first statement where they see we do selection based on merit and not skin color, gender or sexual orientation. Any attempt to introduce gender/race/sexual orientation quotas will promptly be rejected on those basis. There is value in wording it like this beyond "let's do good", reads a comment by “bluecalm” on Hacker News. “The real issue here is that some people get really upset by the word "meritocracy". Clearly, you have never dealt with a white man who is better off than certain women, or people from ethnic minorities, and thinks it's because his being white/male makes him inherently better and invokes "meritocracy" all the freaking time. I have. And I'm a half-Asian male - I can't imagine what it's like to deal with that jerk if I were a woman or an ethnic minority he didn't consider inherently intelligent. When the word "meritocracy" is used to structurally shut down debates of sexism and racism, it is no longer about meritocracy”, reads a comment by “vanderZwan”  on Hacker News. “How is this relevant? If someone is voicing dissent about a culture there's a million barriers they can hide behind instead of facing the allegations ("we're committed to working towards a more diverse and inclusive environment..."). Are we to cower away from any word that can be twisted to justify bad things as well as good things, for fear that people will abuse them? The word "meritocracy" doesn't seem like a line of code specifying an objective action that is to be taken to solve a problem, it describes a vision of how someone would ideally like their company to operate. How about we just call out bad behavior when we see it and not let it pollute the vision? That's what needs to happen, or we'll be locked in this battle until one political group manages to suppress the others (that, by the way, seem to have the exact same end goals they do)”, reads a comment by “nyxxie” on Hacker News. For more information, check out the official Mozilla diversity blog. Mozilla, Internet Society, and web foundation wants G20 to address “techlash” fuelled by security and privacy concerns Mozilla optimizes calls between JavaScript and WebAssembly in Firefox, making it almost as fast as JS to JS calls Mozilla updates Firefox Focus for mobile with new features, revamped design, and Geckoview for Android Mary Meeker, one of the premier Silicon Valley investors, quits Kleiner Perkins to start her own firm

libp2p is a P2P Network stack introduced by the IPFS community. libp2p is capable of discovering other peers and networks without resourcing to centralized registries that enables apps to work offline. In July 2018, Davis Dias explained that the design of a 'location addressed web' is the reason for its fragility. Small errors in its backbone can lead to shutting down of all running applications. Firewalls, routing issues, roaming issue, and network reliability interfere with users having a smooth experience on the web. Thus came a need to re-imagine the network stack. To solve all the above problems, the InterPlanetary File System (IPFS) came into being. It is a decentralized web protocol based on content-addressing, digital signatures, and peer-to-peer distribution. Today, IPFS is used to build completely distributed (and offline-capable!) web-apps which are also available offline. IPFS saves and distributes valuable datasets, and moves billions of files. IPFS spawned several other projects and libp2p is one of them. It enables users to run network applications free from runtime and address services while being independent of their location. libp2p solves the complexity of dealing with numerous protocols in a decentralized environment. It effectively helps users connect with multiple peers using only a single protocol thus paving the way for the next generation of decentralized systems. Libp2p Features #1 Transport Module libp2p enables application developers to pick the modules needed to run their application. These modules vary depending on the runtime they are executing. A libp2p node uses one or more Transports to dial and listen for connections. These transport modules offer a clean interface for dialing and listening which is defined by the interface-transport specification. #2 No prior assigning of ports Before libp2p came into existence, users would assign a listener to a port and then assign ports to special protocols. This was done so that other hosts would know in advance which port to dial. With libp2p users do not have to assign ports beforehand. #3 Encrypted communication To ensure an encrypted connection, libp2p also supports a set of modules that encrypt every communication established. #4 Peer Discovery and Routing A peer discovery module helps libp2p to find peers to connect to. Peer routing finds other peers in the network by intentionally issuing queries, which can be iterative or recursive, until a peer is found. Content routing mechanism is used to find where content lives in the network. Using libp2p in IPFS libp2p is now refactored into its own project so that other users can take advantage of it and be part of its ecosystem as well. It is what provides IPFS and other projects the P2P connectivity, support for multiple platforms and browsers and many other advantages. Users can utilize the libp2p module to create their own libp2p bundle. They can customize their bundles with features and default setup. It also takes into account a user's needs. For example, the team has built a browser working version of libp2p that acts as the network layer of IPFS and leverages browser transports. You can head over to GitHub to check this example. Keep Networks has also demonstrated the use of libp2p. Since participants need to know how to connect to each other, the team has come up with a simple example of peer-to-peer discovery. They have used a few pieces of the libp2p JS library to create nodes that discover and communicate with each other. You can head over to their blog to check out how the example works. Another emerging use for libP2P is in blockchain applications. IPFS is used by blockchains and blockchain applications, and its subprotocols (libp2p, multihash, IPLD) can be extremely useful for blockchain standardization. A good  example of this would be getting the ethereum blockchain in the browser or in a Node.js process using libp2p and running it through ethereum-vm. That being said, there are multiple challenges that developers will encounter while using libP2P for their Blockchain examples. Chris Pacia, the backend developer for OB1, explains how developers can face these challenges in his talk at QCon. With all the buzz around blockchains and decentralized computing these days, libp2p is making its rounds on the internet. For more insights on libp2p, you can visit their official site. Cloudflare’s decentralized vision of the web: InterPlanetary File System (IPFS) Gateway to create distributed websites Tim Berners-Lee plans to decentralize the web with ‘Solid’, an open-source project for “personal empowerment through data” Introducing TLS 1.3, the first major overhaul of the TLS protocol with improved security and speed

Yesterday, Google announced that they will be not be competing for the Pentagon’s cloud-computing contract which is supposedly worth $10 billion. They opted out of bidding for the project named, Joint Enterprise Defense Infrastructure (JEDI) saying the project may conflict with its principles for the ethical use of AI. The JEDI project involves moving massive amounts of Pentagon internal data to a commercially operated secure cloud system. The bidding for this contract began two months ago and closes this week (12th October). CNBC reported in July that Amazon is considered as the number one choice for the contract because it is already providing services for the cloud system used by U.S intelligence agencies. Cloud providers such as IBM, Microsoft, and Oracle are also top-contenders as they have worked with government agencies for many decades. This move could help their chances of winning the decade-long JEDI contract. Why Google has dropped out of this bidding? One of Google’s spokespersons told TechCrunch that the main reason for opting out of this bidding is because it doesn’t align with their AI principles: “While we are working to support the US government with our cloud in many areas, we are not bidding on the JEDI contract because first, we couldn’t be assured that it would align with our AI Principles and second, we determined that there were portions of the contract that were out of scope with our current government certifications.” He further added that: “Had the JEDI contract been open to multiple vendors, we would have submitted a compelling solution for portions of it. Google Cloud believes that a multi-cloud approach is in the best interest of government agencies, because it allows them to choose the right cloud for the right workload. At a time when new technology is constantly becoming available, customers should have the ability to take advantage of that innovation. We will continue to pursue strategic work to help state, local and federal customers modernize their infrastructure and meet their mission critical requirements.” Also, this decision is a result of thousands of Google employees protesting against the company's involvement in another US government project named Project Maven. Earlier this year, some of the Google employees reportedly quit over the company's work on this project. Its employees believed that the U.S. military could weaponize AI and apply the technology towards refining drone strikes and other kinds of lethal attacks. An internal petition was also drafted for Google CEO Sundar Pichai to cancel Project Maven and was signed by over 3,000 employees. After this protest, Google said it would not renew the contract or pursue similar military contracts. Further, Google also formulated its principles for the ethical use of AI. You can read the full story on Bloomberg. Bloomberg says Google, Mastercard covertly track customers’ offline retail habits via a secret million dollar ad deal Ex-googler who quit Google on moral grounds writes to Senate about company’s “Unethical” China censorship plan Google slams Trump’s accusations, asserts its search engine algorithms do not favor any political ideology

bpftrace is a DTrace like tool for troubleshooting kernel problems. It was created about a year ago by Alastair Robertson and the GitHub repository was made public recently. It has plenty of features to relate it to DTrace 2.0. bpftrace bpftrace is an open source high level tracing tool which allows analyzing systems. It is now more competent and built for modern extended Berkeley Packet Filter (eBPF). eBPF is a part of the Linux kernel and is popular in systems engineering. Robertson recently developed struct support, and applied it to tracepoints. Struct support was also applied to kprobes. bpftrace uses existing Linux kernel facilities like eBPF, kprobes, uprobes, tracepoints, and perf_events. It also uses bcc libraries. bpftrace uses a lex/yacc parser internally to convert programs into abstract syntax tree (AST). Then llvm intermediate representation actions are done and finally, then BPF is done. Source: GitHub bpftrace and DTrace bpftrace is a higher-level front end for custom ad-hoc tracing. It can play a similar role as DTrace. There are some things eBPF can do and DTrace can't, one of them being the ability to save and retrieve stack traces as variables. Brendan Gregg, one of the contributors of bpftrace states in his blog: “We've been adding bpftrace features as we need them, not just because DTrace had them. I can think of over a dozen things that DTrace can do that bpftrace currently cannot, including custom aggregation printing, shell arguments, translators, sizeof(), speculative tracing, and forced panics.” A one-liner tutorial and reference guide is available on GitHub for learning bpftrace. For more details and trying bpftrace head on to the GitHub repository and Brendan Gregg’s blog. NVTOP: An htop like monitoring tool for NVIDIA GPUs on Linux LLVM 7.0.0 released with improved optimization and new tools for monitoring Xamarin Test Cloud for API Monitoring [Tutorial]

Yesterday, Google reported a bug discovery in one of the Google+ People APIs, which exposed user’s Google+ profile information such as name, email address, occupation, gender, and age. As per Google’s analysis, the profiles of up to 500,000 Google+ accounts were potentially affected. According to the Wall Street Journal report, “Google opted not to disclose the issue this past spring, in part because of fears that doing so would draw regulatory scrutiny and cause reputational damage.” Google discovered this bug as a part of its Project Strobe, which began in early 2018. Strobe was started with an aim to analyze third-party developer access in Google’s various services and Android. The company says it immediately patched this bug in March 2018 post learning of its existence. The bug provided outside developers potential access to private Google+ profile data between 2015 and March 2018, say internal investigators who discovered and fixed it. Using the API, users can grant access to their profile data, and the public profile information of their friends, to Google+ apps. However, with the bug, the apps also had an access to profile fields even when that data was listed as private and not public. Why were users kept in the dark? Any security breach pertaining to user data exposure should quickly be informed. However, as per the Wall Street Journal report, “A memo reviewed by the Journal prepared by Google’s legal and policy staff and shared with senior executives warned that disclosing the incident would likely trigger ‘immediate regulatory interest’ and invite comparisons to Facebook’s leak of user information to data firm Cambridge Analytica.” In response to the allegations raised on Google, Ben Smith, Vice President of Google’s Engineering team, in his recent blog post mentioned, “Every year, we send millions of notifications to users about privacy and security bugs and issues. Whenever user data may have been affected, we go beyond our legal requirements and apply several criteria focused on our users in determining whether to provide notice.” He also assured that Google’s Privacy & Data Protection Office reviewed the issue. He further added, “looking at the type of data involved, whether we could accurately identify the users to inform, whether there was any evidence of misuse, and whether there were any actions a developer or user could take in response. None of these thresholds were met in this instance.” Ben said that Google found no evidence that any developer was aware of this bug or abusing the API. He also assured that no profile data was misused. Will this delayed bug discovery announcement subject Google to GDPR? The European GDPR (General Data Protection Regulation), which was enforced on 25 May 2018 requires companies to notify regulators of breaches within 72 hours, else the companies would be charged a maximum fine of 2% of world-wide revenue. Al Saikali, a lawyer with Shook, Hardy & Bacon LLP, said, “The information potentially leaked via Google’s API would constitute personal information under GDPR, but because the problem was discovered in March, it wouldn’t have been covered under the European regulation.” He further added, “Google could also face class-action lawsuits over its decision not to disclose the incident. The story here that the plaintiffs will tell is that Google knew something here and hid it. That by itself is enough to make the lawyers salivate.” The Aftermath: Google plans to discontinue Google+ for consumers Ben’s post mentions that over the years, Google+ has not achieved broad consumer or developer adoption, and has seen limited user interaction with apps. Talking about its consumer version, Google+ currently has low usage and engagement--90 percent of Google+ user sessions are less than five seconds. One of the priorities of Project Strobe was to closely review all the APIs associated with Google+ during which it also discovered the bug. Ben mentions, “The review did highlight the significant challenges in creating and maintaining a successful Google+ that meets consumers’ expectations.” Following these challenges and the very low usage of the consumer version of Google+, Google has decided to discontinue Google+ consumer version. This shutdown will take place over the course of the next 10 months, and will conclude in August, next year. However, Google plans to make Google+ available as an enterprise product for companies. Ben states, “We’ve decided to focus on our enterprise efforts and will be launching new features purpose-built for businesses. We will share more information in the coming days.” Other findings of Project Strobe and the actions taken Project Strobe provides a ‘root and branch’ review of third-party developer access to Google account and Android device data and of Google’s philosophy around apps’ data access. The main key finding of this project is the discovery of an exploitable bug built into a core API of Google+ for three years. The other key findings and the actions taken include: The need for having fine-grained control over the data shared with apps For this finding, Google plans to launch more granular Google Account permissions that will show up in individual dialog boxes. Here, instead of seeing all requested permissions in a single screen, apps will have to show the user each requested permission, one at a time, within its own dialog box. Know more about this on Google Developer Blog. Here’s a sample of how this process will look like: Source: Google blog Granting access to user’s Gmail via apps is done with certain use cases in mind For this, Google plans to limit the types of use cases that are permitted. The company is updating their User Data Policy for the consumer Gmail API to limit the apps that may seek permission to access consumer’s Gmail data. Only apps directly enhancing email functionality such as email clients, email backup services and productivity services (e.g., CRM and mail merge services), will be authorized to access this data. Also, these apps will need to agree to new rules for handling Gmail data and will be subject to security assessments. To know more about this action, read the Gmail Developer Blog. Granting SMS, Contacts and Phone permissions to Android apps are done with certain use cases in mind As an action to this finding, Google will limit the apps’ ability to receive call log and SMS permissions on Android devices. Hence, the contact interaction data will no longer be available via the Android Contacts API. Additionally, Google has also provided basic interaction data, for example, a messaging app could show you your most recent contacts. They also plan to remove access to contact interaction data from the Android Contacts API within the next few months. To read more about Project Strobe and the closing down of Google+ in detail, visit Ben Smith Google post. Facebook’s largest security breach in its history leaves 50M user accounts compromised Bloomberg’s Big Hack Exposé says China had microchips on servers for covert surveillance of Big Tech and Big Brother; Big Tech deny supply chain compromise Timehop suffers data breach; 21 million users’ data compromised

Yesterday, Facebook launched two brand new video communication devices. Named Portal and Portal+, these devices let you video call anyone, with more richer, hands-free experiences. The Portal features a 10-inch 1280 x 800 display, while Portal+ features 15-inch 1920 x 1080.  Both devices are powered by Artificial Intelligence. This includes a Smart Camera and a Smart Sound technology. Smart Camera stays with the action and automatically pans and zooms to keep everyone in view. Smart Sound minimizes background noise and enhances the voice of whoever is talking, no matter where they move. Source: Facebook Portal can also be used to call Facebook friends and connections on Messenger even if they don’t have Portal. It also supports group calls of up to seven people at the same time. Portal also offers hands-free voice control with Amazon Alexa built-in which can be used to track sports scores, check the weather, control smart home devices, order groceries, and more.  Facebook has also enabled shared activities in its Portal devices by partnering with Spotify Premium, Pandora, iHeartRadio, Facebook Watch, Food Network, and Newsy. Keeping in mind, it’s security breach that affected 50 million users two weeks ago, Facebook says it has paid a lot of attention to privacy and security features. Per their website, “We designed Portal with tools that give you control: You can completely disable the camera and microphone with a single tap. Portal and Portal+ also come with a camera cover, so you can easily block your camera’s lens at any time and still receive incoming calls and notifications, plus use voice commands. To manage Portal access within your home, you can set a four- to 12-digit passcode to keep the screen locked. Changing the passcode requires your Facebook password. We also want to be upfront about what information Portal collects, help people understand how Facebook will use that information and explain the steps we take to keep it private and secure: Facebook doesn’t listen to, view, or keep the contents of your Portal video calls. In addition, video calls on Portal are encrypted. For added security, Smart Camera and Smart Sound use AI technology that runs locally on Portal, not on Facebook servers. Portal’s camera doesn’t use facial recognition and doesn’t identify who you are. Like other voice-enabled devices, Portal only sends voice commands to Facebook servers after you say, “Hey Portal.” You can delete your Portal’s voice history in your Facebook Activity Log at any time.” In all the above, Facebook seems quite cryptic about audio data. It also doesn’t really explain how it will use the information it collects from users. The voice data is stored on the Facebook server by default, probably to improve the Portal’s understanding on the user’s language quirks and to understand the user’s needs from the data. But it does make one wonder, should this be an opt-in and not an opt-out by default? Another jarring aspect is the need for one’s Facebook password to change the device’s passcode. This just feels like the new devices are yet another way for Facebook to add users to Facebook, not to mention the fact that Facebook just had a data breach on its site, the repercussions of which they are still investigating. In an interesting poll conducted by Dr. Jen Golbeck, Professor at UMD, on Twitter, over 63% of respondents said that they will not trust Facebook to responsibly operate a surveillance device in their home. https://twitter.com/jengolbeck/status/1049343277110054912 Read more about the devices on Facebook’s announcement. Facebook Dating app to release as a test version in Colombia. Facebook’s Glow, a machine learning compiler, to be supported by Intel, Qualcomm and others How Facebook is advancing artificial intelligence [Video]

Yesterday, Mozilla announced that in the latest version of Firefox Beta, calls between JS and WebAssembly are faster than non-inlined JS to JS function calls. They have made these optimizations keeping two aspects of engine’s work in mind: reducing bookkeeping and cutting out intermediaries. How they made WebAssembly function calls faster They have optimized the calls in both directions, that is, from JavaScript to WebAssembly and WebAssembly to JavaScript with their recent work in Firefox. All these optimizations have been done to make the engine’s work easier. The improvements fall into two groups: Reducing bookkeeping: This means getting rid of unnecessary work to organize stack frames Cutting out intermediaries: This means taking the most direct path between functions How they optimized WebAssembly to JavaScript calls The browser engine has to deal with two different kinds of languages while going through your code even if the code is all written in JavaScript: bytecode and machine code. The engine needs to be able to go back and forth between these two languages. When it does these jumps, it needs to have some information in place, like the place from where it needs to resume. The engine also must separate the frames that it needs. To organize its work, the engine gets a folder and puts this information in it. When the Firefox developers first added WebAssembly support, they had a different type of folder for it. So even though JIT-ed JavaScript code and WebAssembly code were both compiled and speaking machine language, it was treated as if they were speaking different languages. This was unnecessarily costly in two ways: An unnecessary folder is created which adds up setup and teardown costs It requires trampolining through C++ to create the folder and do other setup They fixed this by generalizing the code to use the same folder for both JIT-ed JavaScript and WebAssembly. This made calls from WebAssembly to JS almost as fast as JS to JS calls. How they optimized JavaScript to WebAssembly calls JavaScript and WebAssembly use different customs even if they are speaking the same language. For instance, to handle dynamic types, JavaScript uses something called boxing. As JavaScript doesn’t have explicit types, they need to be figured out at runtime. To keep track of the types of values, the engine attaches a tag to the value. This turns one simple operation into four operations. This is the reason why WebAssembly expects parameters to be unboxed and doesn’t box its return values. Since it is statically typed, it doesn’t need to add this overhead. So, before the engine gives the parameters to the WebAssembly function, the engine needs to unbox the values and put them in registers. It has to go through C++ again to prepare the values when going from JS to WebAssembly. Going to this intermediary step is a huge cost, especially for something that’s not that complicated. To solve this, they took the code that C++ was running and made it directly callable from JIT code. So, when the engine goes from JavaScript to WebAssembly, the entry stub unboxes the values and places them in the right place. Along with these calls, they have also optimized monomorphic and built-in calls. To understand the optimizations well, check out Lin Clark’s official announcement on Mozilla’s website. Mozilla updates Firefox Focus for mobile with new features, revamped design, and Geckoview for Android Mozilla releases Firefox 62.0 with better scrolling on Android, a dark theme on macOS, and more Mozilla, Internet Society, and web foundation wants G20 to address “techlash” fuelled by security and privacy concerns

People started using htop when the top just didn’t provide enough information. Now there is NVTOP, a tool that looks similar to htop but displays the process information loaded on your NVIDIA GPU. It works on Linux systems and displays detailed information about processes, memory used, which GPU and also displays the total GPU and memory usage. The first version of this tool was released in July last year. The latest change made the process list and command options scrollable. Some of the features of NVTOP are: Sorting by column To Select / Ignore a specific GPU by ID To kill selected process Monochrome option Yes, it has multi GPU support and can display the running processes from all of your GPUs. The information printed out looks like the following, and is similar to something htop would display. Source: GitHub There is also a manual page to give some guidance in using NVTOP. It can be accessed with this command: man nvtop There are OS specific installation steps on GitHub for Ubuntu/Debian, Fedora/RedHat/CentOS, OpenSUSE, and Arch Linux. Requirements There are two libraries needed to build and run NVTOP: The NVIDIA Management Library (NVML) for querying GPU information. The ncurses library for the user interface and make it colorful. Supported GPUs The NVTOP tool works only for NVIDIA GPUs and runs on Linux systems. One of the dependencies is the NVML library which does not support some queries from GPUs before the Kepler microarchitecture. That is anything before GeForce 600 series, GeForce 700 series, or GeForce 800M wouldn’t likely work. For AMD users, there is a tool called radeontop. The tool is provided under the GPLV3 license. For more details, head on to the NVTOP GitHub repository. NVIDIA leads the AI hardware race. But which of its GPUs should you use for deep learning? NVIDIA announces pre-orders for the Jetson Xavier Developer Kit, an AI chip for autonomous machines, at $2,499 NVIDIA open sources its material definition language, MDL SDK

Microsoft announced a new game streaming project for Xbox yesterday, namely, Project xCloud. Microsoft is not the only one who has rolled out their game streaming service, Google announced a new experimental game streaming service, namely, Project Stream, last week, that lets the users play Assassin’s Creed in Chrome among other things. Project xCloud is a game streaming technology that enables existing Xbox users to play on any device that they want to without being limited to a particular device. This means gamers can choose either console or PC depending on their preference. Mobile-only players can also access the worlds, characters and immersive stories that they haven’t been able to experience before on a mobile. Project xCloud provides gamers with plenty of options and does so without altering the performance and experience of the game. “Our goal with Project xCloud is to deliver a quality experience for all gamers on all devices that’s consistent with the speed and high-fidelity gamers experience and expect on their PCs and consoles”, mentions Kareem Choudhry, Corporate VP, Gaming Cloud, Microsoft on the announcement page.” Microsoft Microsoft admits that Cloud game-streaming is a “multi-faceted, complex challenge”. In order to provide high-quality experience across different devices, it is crucial to watch out for obstacles, such as low-latency video streamed remotely, and support for a large, multi-user network. “With our nearly 40 years of gaming experience starting with PC, as well as our deep experience of being a platform company, we are equipped to address the complex challenge of cloud game-streaming”, mentions Choudhary. Microsoft had been working on Project xCloud over the past few years and aims to bring the public trials of this project in 2019. Moreover, Microsoft wants to make it easy for developers to bring their content to Project xCloud and deliver an amazing experience to the existing Xbox players. Developers will be able to deploy and dramatically scale to hundreds of millions of users across all devices on Project xCloud with no additional work. Microsoft has data centers in 54 Azure regions and services available in 140 countries, meaning that Azure has the scale to offer great gaming experience for players worldwide, regardless of their location. “We’ve enabled compatibility with existing and future Xbox games by building out custom hardware for our datacenters that leverages our years of console and platform experience. We’ve architected a new customizable blade that can host the component parts of multiple Xbox One consoles, as well as the associated infrastructure supporting it. We will scale those custom blades in datacenters across Azure regions over time”, writes Choudhry. Project xCloud was tested yesterday by Microsoft. The test was carried out on different devices (mobile phones, tablets) paired with an Xbox Wireless Controller through Bluetooth. As of now, the test experience is running at 10 megabits per second. Microsoft is also developing a new, game-specific touch input overlay that offers a maximum response in a minimal footprint to players wanting to play without a controller. Additionally, Project xCloud makes game streaming possible on 4G networks. “We are looking forward to learning with you during our public trials next year and sharing more details as we continue on this journey to the future of gaming with you at the center”, says Choudhary. For more information, check out the official Microsoft blog. Microsoft open sources Infer.NET, it’s popular model-based machine learning framework Microsoft announces new Surface devices to enhance user productivity, with style and elegance Microsoft’s new neural text-to-speech service lets machines speak like people

On 8th October, at it’s 'Fall Desktop Launch Event', Intel unveiled the 9th-generation Core i9-9900K, i7-9700K, and i5-9600K processors for desktops. With an aim to deliver ‘the best gaming performance’ in the word, the processors also come with fixes for the much controversial  Specter, Meltdown, and L1TF vulnerabilities. Major features of this launch include, #1 Security fixes for Specter, Meltdown, and LITF Faults In March 2018, Intel announced that they would be adding hardware protection to forthcoming CPUs protecting users against some of the processor's security flaws. These 'protective walls' added in the hardware would keep malicious code in a physically different location from areas of the CPU were speculative execution is taking place. Intel kept its word by announcing hardware mitigations in the 9th Gen CPU’s for Spectre/Meltdown. Former Intel CEO Brian Krzanich stated in a press release, "We have redesigned parts of the processor to introduce new levels of protection through partitioning that will protect against both Variants 2 and 3. Think of this partitioning as additional “protective walls” between applications and user privilege levels to create an obstacle for bad actors." It has not been detailed what specific hardware changes were made to add protection. It was noted that the previous software and microcode protections added would cause a performance hit on older CPUs. These new CPUs are powerful enough that any performance hit caused by these protections should not be noticeable. #2 Forgoing HyperThreading Intel is forgoing HyperThreading on some of the Core i9 parts. This will partly help make the product stack more linear. This could also possibly help mitigate one of the side-channel attacks that can occur when HyperThreading is in action. Disabling HyperThreading on the volume production chips, ensures that every thread on that chip is not competing for per-core resources. #3 Hardware Specifications Source: AnandTech Core i9-9900K The  Core i9-9900K processor is designed to deliver the best gaming performance in the world. Users can enable up to 220 FPS on Rainbow Six: Siege, Fortnite, Counter-Strike: Global Offensive and PlayerUnknown Battlegrounds. It comes with8 cores, 16 threads and a base frequency of 3.6GHz which can be boosted up to 5.0GHz. This processor is aimed at desktop-based enthusiasts and with a dual-channel DDR4 and up to 40 PCIe lanes. The i9-9900K is based off Intel’s 14nm process. Hyperthreading is an added bonus in this processor. Core i7-9700K The i7-9700K comes with 8 cores and 8 threads. With a  base clock speed is of 3.6 GHz (which can be boosted to 4.9 GHz on all cores), the processor comes without hyperthreading.  It can turbo up to 4.9 GHz only on a single core. The i7-9700K is meant to be the direct upgrade over the Core i7-8700K. While both chips have the same Coffee Lake microarchitecture, the 9700K has two more cores and slightly better turbo performance. That being said, it has less L3 cache per core at only 1.5MB per core. Core i5-9600K The  i5-9600K is clocked at a base frequency of 3.7 GHz and can be boosted up to 4.6 GHz. With 6 cores and 6 threads, it comes without Hyperthreading. This processor is really similar to the Core i5 of the previous generation, but with an added frequency for better performance. It would be interesting to see how these new processors will help in mitigating security flaws without impacting their performance. For detailed information on each of the processors, you can head over to AnandTech. You could also check out BleepingComputer for additional insights. NetSpectre attack exploits data from CPU memory Intel faces backlash on Microcode Patches after it prohibited Benchmarking or Comparison Meet ‘Foreshadow’: The L1 Terminal Fault in Intel’s chips

At the end of last week, on October 5, Elastic - the company behind the hugely popular ElasticSearch search tool - went public. And it looks like the move public paid off. By the end of its first day as a public company, shares in Elastic had doubled in price. At the start of the day it was around $36 per share - by the close of trading it has leapt to $70. This meant its market cap had risen from $2.5 billion to $4.9 billion. We probably shouldn't be surprised. As Fortune pointed out over the weekend, the demand for shares in Elastic was already getting pretty hot towards the end of September. "It originally filed to sell the stock at $26 to $29 apiece on Sept. 24, while Thursday’s pricing was higher even than banks managing the sale were expecting." https://twitter.com/elastic/status/1048302021143670786 Elastic's journey to the stock market ElasticSearch was first released back in February 2010. The tool contained the imprint of Elastic's mission - to make powerful search accessible to modern businesses. And it has certainly done just that. Today Elastic powers immensely popular apps like Uber and Tinder, as well as giving logging and processing power to companies like Cisco where big data has become the norm. According to founder Shay Banon, writing in a blog post published on Friday, Elastic has seen "more than 350 million product downloads, a... community of more than 100,000 developers, and more than 5,500 customers." What next for Elastic? Banon insists that "as a public company, [Elastic] will continue doing the things that have made us Elastic." This includes continued investment in the developer communities that have grown up around Elastic's products, new features to products and working with customers to adapt to changing trends in software infrastructure - so Elastic's products can be deployed anywhere. Business as usual might well be a recipe for success for Elastic. Investors will be hoping that the organization continues to deliver on its mission, as demand for better, faster search isn't going to end any time soon.

The Windows 10 October update was available for download around the time of the Surface event last week. While the update brought features like Your Phone App and Windows Timeline, users also experienced massive file deleting from their systems. Microsoft had excluded the update from some devices due to compatibility issues with newer processors. The issue was reported by users in the early stages before mass rollout. Users could manually download and install the Windows October 2018 Update from October 2. Rollout was to be pushed October 9 for Patch Tuesday. Microsoft recommends contacting their customer support if the update has deleted your files. The support site advices: “If you have manually downloaded the Windows 10 October 2018 Update installation media, please don’t install it and wait until new media is available.” As of now, it is not known how many users faced this issue. Windows updates are not known to be smooth, causing some issues and errors. But it is unusual that an issue of this magnitude was not detected in Microsoft’s testing of the Windows update. Earlier this year, Microsoft had delayed the Windows 10 April 2018 because of Blue Screen of Death issues. But the issues in that update were rectified before the update reached regular users. Fortunately, this update wasn’t mass rolled out and the issue was detected in an early stage. This serves as a reminder to users to create a backup of important files before an OS update. When Microsoft continues mass rollout of this update, the issue will be fixed, but it is safe to backup your data in any case. The official support page states: “We have paused the rollout of the Windows 10 October 2018 Update (version 1809) for all users as we investigate isolated reports of users missing some files after updating.” There are comments on the support page, where users are stating the problem. For more details visit the Microsoft support website. Microsoft Your Phone: Mirror your Android phone apps on Windows What’s new in the Windows 10 SDK Preview Build 17704 Microsoft Cloud Services get GDPR Enhancements