In the News
A new AI tool to fight the coronavirus
A coalition of AI groups is forming to produce a comprehensive data source on the coronavirus pandemic for policymakers and health care leaders.
Sponsor
Find & Address Cloud Infrastructure Risk In Minutes. View Demo and Try for Free!
Detect malware, misconfigurations, lateral movement risk, secret keys & risky passwords. Instant-On. No agents. No overlooked assets. No DevOps headaches. No performance hit. Connect your first cloud account in minutes & see for yourself.
In The News
Why are Artificial Intelligence systems biased?
A machine-learned AI system used to assess recidivism risks in Broward County, Fla., often gave higher risk scores to African Americans than to whites, even when the latter had criminal records.
Watch artificial intelligence learn to simulate sloppy mixtures of water, sand, and ‘goop’
When scientists or special effects wizards want to simulate a flood or visualize an asteroid impact, they turn to programs called physics engines.
Applied use cases
Graphcore claims its M2000 AI computer hits 1 petaflop
AI accelerators like the GC200 are a type of specialized hardware designed to speed up AI applications, particularly artificial neural networks, deep learning, and machine learning.
A beginner’s guide to how machines learn
Introduction Once you get into artificial intelligence and machine learning, there’s no way to avoid three terms: Supervised learning Unsupervised learning Reinforcement learning
Alexa, go to the kitchen and fetch me a snack
MIT researchers have developed a representation of spatial perception for robots that is modeled after the way humans perceive and navigate the world.
Ethics
What our students have been working on during lockdown – Elsa
One of the most common exercises for ML beginners is a Naïve Bayes spam email classifier. To gain a more detailed understanding of how ML techniques can be applied in interdisciplinary research, I decided to work part time as a research annotator at the Alan Turing Institute.
Q&A: The Data Delusion
He highlights the need to update legal privacy standards to be more reflective of the harms incurred through collective data analysis, as opposed to individual privacy violations.
An invisible hand: Patients aren’t being told about the AI systems advising their care
At a growing number of prominent hospitals and clinics around the country, clinicians are turning to AI-powered decision support tools — many of them unproven — to help predict whether hospitalized patients are likely to develop complications or deteriorate, whether they’re at risk of readmission,...
Robotics
Roomba and the role of future robots
Today, the house-cleaning Roomba seems almost ubiquitous, but in a recent essay, its inventor, Joe Jones, recalls his wrong prediction in the 1980s that “in three to five years, robots will be everywhere doing all sorts of jobs.”
Letting robots manipulate cables
For humans, it can be challenging to manipulate thin flexible objects like ropes, wires, or cables. As a cable slides between the fingers, its shape is constantly changing, and the robot’s fingers must be constantly sensing and adjusting the cable’s position and motion.
This Robotic Chemist Does Over 600 Experiments a Week and Learns From Its Own Work
Now though, researchers at the University of Liverpool in the UK have created a mobile robot that can carry out experiments using standard lab equipment designed for humans and can make decisions on the fly about what experiments it should do next based on its previous results.
Research
Gradient descent for wide two-layer neural networks – II: Generalization and implicit bias
Our line of reasoning however completely falls apart without such a regularization: if the objective function depends on the predictor only via its values on the training set, being a minimizer does not guarantee anything about generalization outside of the training set
Productionizing machine learning models, one thoughtful change at a time with Josh Tobin
Josh Tobin, former researcher at OpenAI and creator of Full Stack Deep Learning talks about professionalizing ML workflows for the real world, his work with the Robotics team and FSDL.
One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Control.
Reinforcement learning is typically concerned with learning control policies tailored to a particular agent. We investigate whether there exists a single global policy that can generalize to control a wide variety of agent morphologies