What's important is that you have a faith in people, that they're basically good and smart, and if you give them tools, they'll do wonderful things with them. In the nearly 20 years since I started medical school, I've seen the practice of medicine undergo a wholesale technological transformation. Take medical records as a simple example. I am 100% certain that today's medical students are much slower walkers than me. Because the days of sprinting on rounds to get ahead of the white coat phalanx, pull down a cabinet and open a three-ring binder chart to the next blank page before the intern reaches the door ended a decade ago.

The figure above illustrates the method:(a) Goal-conditioned RL often fails to reach distant goals, but can successfully reach the goal if starting nearby (inside the green region).

Artificial intelligence is reshaping how we live, learn, and work, and this past fall, MIT undergraduates got to explore and build on some of the tools and coming out of research labs at MIT. Through the Undergraduate Research Opportunities Program (UROP), students worked with researchers at the MIT Quest for Intelligence and elsewhere on projects to improve AI literacy and K-12 education, understand face recognition and how the brain forms new memories, and speed up tedious tasks like cataloging new library material. Six projects are featured below. Nicole Thumma met her first robot when she was 5, at a museum. "It was incredible that I could have a conversation, even a simple conversation, with this machine," she says.

University of Oxford scientists led research that used artificial intelligence to generate accurate machine learning emulator algorithms for accelerating simulations billions of times, for all scientific disciplines. Researchers led by the University of Oxford in the U.K. used artificial intelligence to generate accurate machine learning emulator algorithms for accelerating simulations billions of times, for all scientific disciplines. The neural network-based emulators absorb the inputs and outputs of a full simulation, seeking patterns and learning to guess what the model would do with new inputs while avoiding the need to run the full simulation many times. The Deep Emulator Network Search (DENSE) method randomly inserts computation layers between network inputs and outputs and trains the system with the limited data, so added layers that improve performance are more likely to end up in future variations. DENSE-produced emulators for 10 simulations in physics, astronomy, geology, and climate science were 100,000 to 2 billion times faster than the models with the addition of specialized graphical processing chips--and were highly accurate.

We study the online learning problem of a bidder who participates in repeated auctions. With the goal of maximizing his T-period payoff, the bidder determines the optimal allocation of his budget among his bids for $K$ goods at each period. As a bidding strategy, we propose a polynomial-time algorithm, inspired by the dynamic programming approach to the knapsack problem. The proposed algorithm, referred to as dynamic programming on discrete set (DPDS), achieves a regret order of $O(\sqrt{T\log{T}})$. By showing that the regret is lower bounded by $\Omega(\sqrt{T})$ for any strategy, we conclude that DPDS is order optimal up to a $\sqrt{\log{T}}$ term.

We consider the optimization of cost functionals on manifolds and derive a variational approach to accelerated methods on manifolds. We demonstrate the methodology on the infinite-dimensional manifold of diffeomorphisms, motivated by registration problems in computer vision. We build on the variational approach to accelerated optimization by Wibisono, Wilson and Jordan, which applies in finite dimensions, and generalize that approach to infinite dimensional manifolds. We derive the continuum evolution equations, which are partial differential equations (PDE), and relate them to simple mechanical principles. Our approach can also be viewed as a generalization of the $L 2$ optimal mass transport problem.

We formulate the problem of metric learning for k nearest neighbor classification as a large margin structured prediction problem, with a latent variable representing the choice of neighbors and the task loss directly corresponding to classification error. We describe an efficient algorithm for exact loss augmented inference,and a fast gradient descent algorithm for learning in this model. The objective drives the metric to establish neighborhood boundaries that benefit the true class labels for the training points. Our approach, reminiscent of gerrymandering (redrawing of political boundaries to provide advantage to certain parties), is more direct in its handling of optimizing classification accuracy than those previously proposed. In experiments on a variety of data sets our method is shown to achieve excellent results compared to current state of the art in metric learning.

In this paper, we propose a winner-take-all method for learning hierarchical sparse representations in an unsupervised fashion. We first introduce fully-connected winner-take-all autoencoders which use mini-batch statistics to directly enforce a lifetime sparsity in the activations of the hidden units. We then propose the convolutional winner-take-all autoencoder which combines the benefits of convolutional architectures and autoencoders for learning shift-invariant sparse representations. We describe a way to train convolutional autoencoders layer by layer, where in addition to lifetime sparsity, a spatial sparsity within each feature map is achieved using winner-take-all activation functions. We will show that winner-take-all autoencoders can be used to to learn deep sparse representations from the MNIST, CIFAR-10, ImageNet, Street View House Numbers and Toronto Face datasets, and achieve competitive classification performance.

SafetyPay's platform allows non-card holders and fraud-wary consumers to participate in the online marketplace via bank transfer or cash without sharing their information online. The platform opens the door for e-commerce merchants to tap into a larger consumer base by accepting alternative forms of payment. Meanwhile, Feedzai monitors patterns in payment transaction activity and compares against a customer's historical data to authenticate transactions. With a shared goal of making banking and commerce safe, the partnership with Feedzai enhances SafetyPay's security, harnessing AI to protect customers across borders from fraudulent risks in real-time. "Secure payments have been a core focus for us since SafetyPay was founded more than a decade ago," said Gustavo Ruiz Moya, CEO, SafetyPay.

A new "text-generating system" created by the brains behind Massachusetts Institute of Technology may be the beginning of the end for all human editing jobs. The system, announced in a press release Wednesday, is able to rummage through the millions of Wikipedia pages, sniff around for outdated data, and replace it with the most recent information available on the internet in a "human-like" style -- thus making the need for real, hot-blooded editors basically obsolete.