Members of the SARA team are pictured in Duffield Hall. From left: Duncan Sutherland, Ph.D. student in materials science and engineering; Carla Gomes, professor of computer science; Mike Thompson, professor of materials science and engineering; and Sebastian Ament, Ph.D. student in computer science. When a master chef develops a new cake recipe, she doesn't try every conceivable combination of ingredients to see which one works best. The chef uses prior baking knowledge and basic principles to more efficiently search for that winning formula. Materials scientists use a similar method in searching for novel materials with unique properties in fields such as renewable energy and microelectronics.

When a master chef develops a new cake recipe, she doesn't try every conceivable combination of ingredients to see which one works best. The chef uses prior baking knowledge and basic principles to more efficiently search for that winning formula. Materials scientists use a similar method in searching for novel materials with unique properties in fields such as renewable energy and microelectronics. And a new artificial intelligence tool developed by Cornell researchers promises to rapidly explore and identify what it takes to "whip up" new materials. SARA (the Scientific Autonomous Reasoning Agent) integrates robotic materials synthesis and characterization, along with a hierarchy of artificial intelligence and active learning methods, to efficiently reveal the structure of complex processing phase diagrams, making materials discovery vastly quicker.

When a master chef develops a new cake recipe, she doesn't try every conceivable combination of ingredients to see which one works best. The chef uses prior baking knowledge and basic principles to more efficiently search for that winning formula. Materials scientists use a similar method in searching for novel materials with unique properties in fields such as renewable energy and microelectronics. And a new artificial intelligence tool developed by Cornell researchers promises to rapidly explore and identify what it takes to "whip up" new materials. SARA (the Scientific Autonomous Reasoning Agent) integrates robotic materials synthesis and characterization, along with a hierarchy of artificial intelligence and active learning methods, to efficiently reveal the structure of complex processing phase diagrams, making materials discovery vastly quicker.

It is well known that (stochastic) gradient descent has an implicit bias towards wide minima. In deep neural network training, this mechanism serves to screen out minima. However, the precise effect that this has on the trained network is not yet fully understood. In this paper, we characterize the wide minima in linear neural networks trained with a quadratic loss. First, we show that linear ResNets with zero initialization necessarily converge to the widest of all minima. We then prove that these minima correspond to nearly balanced networks whereby the gain from the input to any intermediate representation does not change drastically from one layer to the next. Finally, we show that consecutive layers in wide minima solutions are coupled. That is, one of the left singular vectors of each weight matrix, equals one of the right singular vectors of the next matrix. This forms a distinct path from input to output, that, as we show, is dedicated to the signal that experiences the largest gain end-to-end. Experiments indicate that these properties are characteristic of both linear and nonlinear models trained in practice.