Are ANNs capable of Few-Shot Learning?

January 9, 2021

Speaker: Som V Tambe, an Undergraduate student at the Dept. of Aerospace Engineering, IIT Kanpur.
Title: Are ANNs capable of Few-Shot Learning?

Despite remarkable advances in artificial intelligence and machine learning, machine systems have lagged behind Human Learning in two aspects. First, people can learn a new concept from just one or a handful of examples, whereas standard algorithms in machine learning require tens or hundreds of examples to perform similarly. Second, people learn richer representations than machines do, even for simple concepts, using them for a wider range of functions. Past efforts to counter these problems include the Bayesian Program Learning, which follows the idea of “how humans do one-shot classification”. The Bayesian Program Learning paper was remarkable but seems to have hand-engineered various parts of the model.

Our talk focuses on developing more human-like algorithms like One-Shot Learning. This talk will involve Brendan Lake’s paper, as well as other related literature. The speaker shall then discuss the studies he conducted with his mentor, Shashikant Gupta, which provides a completely deep-learning based approach to tackle the Omniglot dataset, and the progress they have achieved with the project.

Slide: Link
Video: Youtube