Last week, the ML world was all gaga about the launch of GitHub Copilot, an AI-based tool that helps programmers write better code. Copilot is based on OpenAI Codex, an AI system trained on a dataset made up of a sizable chunk of public source code at Github.
The announcement of GitHub Copilot kicked up a storm in social media. The initial reaction has been largely positive, with many people calling the coding assistant a game-changer.
However, few questions have been raised: Since the tool is trained on publicly available code repositories–most of which are licensed and under copyright protection–what happens when the tool reproduces these code snippets? Is it legal? Can parent organisations –Microsoft, OpenAI and GitHub– monetise this tool even if it is trained on free and open-source code?
India’s 2nd cloud region: Delhi NCR
Join us at the official opening of the second Google Cloud region.
Google’s Second Cloud Region In India: Implications & Competition
Google cloud regions are a deployment area for Google Cloud Platform resources. Each of these regions consists of zones, and every zone is designated as the single failure domain within that region. Most regions have three or more such zones.
Currently, Google has 24 of these cloud regions and 73 zones across 17 countries. In 2020, four of these regions were launched.
The Delhi cloud region will be officially launched next week and will have three zones to protect against service disruptions.
Softbank Puts Pepper On Hold
Seven years ago, Japanese conglomerate SoftBank launched Pepper, the first humanoid robot that can read emotions. Last year, the company paused its production and would only make the robot ‘when it is needed’. Softbank is also downsizing its global robotics operation in France.
So, Why did Pepper fail?
How To Optimise Deep Learning Models
Increasing number of parameters, latency, resources required to train etc have made working with deep learning tricky. Google researchers, in an extensive survey, have found common challenging areas for deep learning practitioners and suggested key checkpoints to mitigate these challenges.
Most of these challenges boil down to lack of efficiency. Practitioners should aim to achieve pareto-optimality i.e. any model we choose should have the best of tradeoffs. And, one can develop a pareto-optimal model using the following mental model.
Featured Video | AI Startup Story - Elon Musk experiments with Neuralink
How Uber Implements CI/CD Of ML Models
Uber in its latest blog post highlighted various pain points, alongside explaining the solution implementation of continuous integration (CI) and continuous deployment (CD) of machine learning models as a solution.
Uber addressed the following MLOps challenges, such as behaviour changes with new releases, dependency changes and service build script changes by employing a three-stage strategy for validating and deploying the latest binary of the real-time prediction service.
Hands-on Guides for ML Developers
PEOPLE & STARTUPS
CVPR 2021 Best Paper Award
Andreas Geiger and Michael Niemeyer from Max Planck Institute for Intelligent Systems and the University of Tubingen have won the best paper award at CVPR 2021 (Conference on Computer Vision and Pattern Recognition).
Their paper titled — ‘ GIRAFFE : Representing Scenes as Compositional Generative Neural Feature Fields’, explored generating new images and controlling what will appear, the objects and their positions and orientations, the background, etc. Using a modified GAN architecture, they can even move objects in the image without affecting the background or the other objects.
Is AI A Good Judge Of Cause & Effect?
Incorporating insights of psychology research into algorithms is tricky as the former is not exactly a quantifiable metric. But, it can be quite useful as algorithms are venturing into a world full of “trolley problems” in the form of self-driving cars and medical diagnosis.
Tobias Gerstenbeg, assistant professor of psychology at Stanford, believes that by providing a more quantitative characterisation of a theory of human behavior and instantiating that in a computer program, we can make it easier for a computer scientist to incorporate such insights into an AI system. Gerstenbeg and his colleagues at Stanford have developed a computational model to understand how humans judge causation in dynamic physical situations.
BOTTOM OF THE NEWS
Here's what all happened last week.