This week the machine learning community had their handsful with OpenAI’s new toy GPT-3. Many enthusiasts applied the model for various innovative uses and few even started startups that work on GPT-3. Apart from this, there are also reports of the quarterly earnings, which saw Microsoft performing well, especially in the cloud segment. Know what else has happened in this week’s top AI news.
GitHub Goes to Arctic
In a recent development, GitHub moved 21TB of its open-source code and repositories in the form of digital photosensitive archival film into Arctic Code Vault, Svalbard. The boxes of reels are stored in hundreds of meters of permafrost and can last for 1000 years. Done in collaboration with their archive partners, Piql. This initiative, ‘GitHub Archive Program’, aims to preserve the open-source software for future generations.
India To Get Quantum Cloud Service
D-Wave Systems, a Canadian quantum computing company announced the expansion of its Leap cloud access and quantum application environment to India and Australia. The company claims that now users in these countries will have real-time access to a commercial quantum computer. In addition to access, Leap offers free developer plans, teaching and learning tools, code samples, demos and an emerging quantum community to help developers, forward-thinking business and researchers get started building and deploying quantum applications.
The race to democratise has made MLaaS a lucrative business model. The result is, today, there are multiple APIs offering similar services. This again, can be challenging. Addressing this issue and to establish a hassle-free ML ecosystem, a group of researchers from Stanford University, introduced a predictive framework called FrugalML that assists the users in switching between APIs in a smart manner. The researchers have detailed about their new framework in a paper titled, ‘To Call or Not to Call?’
The results show that FrugalML leads to more than 50% cost reduction when using APIs from Google, Microsoft and Face++ for a facial emotion recognition task. Whereas, experiments on FER+ dataset showed that only 33% cost is needed to achieve accuracies that match those of Microsoft API.
The authors posit that the performance of Frugal ML is likely because the base service’s quality score is highly correlated to its prediction accuracy, and their framework only needs to call expensive services for a few difficult data points and relies on the cheaper base services for the relatively easy data points.
Azure Reaps Huge Profit For Microsoft
Microsoft on Wednesday, reported earnings for its fourth fiscal quarter of 2020, including revenue of $38.0 billion, net income of $11.2 billion, and earnings per share of $1.46 (compared to revenue of $33.7 billion, net income of $13.2 billion, and earnings per share of $1.71 in Q4 2019). All three of the company’s operating groups saw year-over-year growth.
“ Organizations that build their own digital capability will recover faster and emerge from this crisis stronger.”-Satya Nadella, CEO, Microsoft
Revenue in Intelligent Cloud was $13.4 billion and increased 17% (up 19% in constant currency). The server products and cloud services revenue increased 19% (up 21% in constant currency) driven by Azure revenue growth of 47% (up 50% in constant currency). Whereas, the enterprise Services revenue was relatively unchanged (up 2% in constant currency).
Python Tops Charts
In a recent survey conducted by IEEE Spectrum, it was found that Python has exerted sheer dominance over its contemporaries Java and C. The organisers have devised 11 metrics to check the popularity of 55 languages. “One interpretation of Python’s high ranking is that its metrics are inflated by its increasing use as a teaching language: Students are simply asking and searching for the answers to the same elementary questions over and over,” stated IEEE in their blog. The rose in Python’s popularity also coincides with that of fields such as machine learning, which have been increasingly introducing libraries and frameworks that encourage Python users. Given the recent trends, it looks like there are no roadblocks in sight for Python.
The New GPT-3 Obsession
GPT-3, the world’s largest NLP model, which was released by OpenAI last month became quite popular. From generating codes to believable stories, this model has been put to use for a wide range of applications.
“Generative models can display both overt and diffuse harmful outputs, such as racist, sexist, or otherwise pernicious language. This is an industry-wide issue, making it easy for individual organizations to abdicate or defer responsibility. OpenAI will not.”
The popularity rose so high that one of the founders of OpenAI, Sam Altman, had to put out a tweet warning how GPT-3 is still far from being perfect. While the OpenAI team is jubilant of this rapid adoption, have listed a set of guidelines explaining how they would be working on making GPT-3 more reliable in the coming days.
Auto Discover RL Algorithms Now
DeepMind researchers released a paper that details a meta learning approach that would allow the researchers to automate the discovery of reinforcement learning algorithms, which have been manual so far. The paper claims that the generated algorithms performed well in video games such as Atari.
“The proposed approach has the potential to dramatically accelerate the process of discovering new reinforcement learning algorithms by automating the process of discovery in a data-driven way,” wrote the researchers.
Drivers Sue Uber
According to VICE reports, Four United Kingdom Uber drivers launched a lawsuit on Monday to gain access to Uber’s algorithms through Europe’s General Data Protection Regulation (GDPR).
The union representing the drivers said they’re seeking to gain a deeper understanding of the algorithms that underpin Uber’s “automated decision-making” system. This level of transparency, the union said, is needed to establish the level of “management control” Uber exerts on its drivers, allow them to calculate their true wages and benchmark themselves against other drivers, and help them build “collective bargaining power.”
The information asymmetry that allows Uber to selectively share data in forms that paint it in a favorable light—usually by obscuring negative outcomes like “dead mileage” or arbitrary deactivation. The case is being heard in Amsterdam and the outcome can severely impact the way Uber and other ride hailing companies do their business.
NVIDIA To Build AI Supercomputer
The University of Florida on Wednesday has announced a public-private partnership with NVIDIA that will catapult UF’s research strength to address some of the world’s most formidable challenges, create unprecedented access to AI training and tools for underrepresented communities, and build momentum for transforming the future of the workforce.
The initiative is anchored by a $50 million gift — $25 million from UF alumnus Chris Malachowsky and $25 million in hardware, software, training and services from NVIDIA, the Silicon Valley-based technology company he co founded and a world leader in AI and accelerated computing.
Along with an additional $20 million investment from UF, the initiative will create an AI-centric data center that houses the world’s fastest AI supercomputer in higher education. Working closely with NVIDIA, UF will boost the capabilities of its existing supercomputer.