30 citations · 63 across the 10 of their papers we have counts for
5 papers · 1 filter
Open-Ended Content-Style Recombination Via Leakage Filtering
Karl Ridgeway, Michael C. Mozer
We consider visual domains in which a class label specifies the content of an image, and class-irrelevant properties that differentiate instances constitute the style. We present a…
Sparse Attentive Backtracking: Temporal CreditAssignment Through Reminding
Nan Rosemary Ke, Anirudh Goyal, Olexa Bilaniuk +4
Learning long-term dependencies in extended temporal sequences requires credit assignment to events far back in the past. The most common method for training recurrent neural netwo…
State-Denoised Recurrent Neural Networks
Michael C. Mozer, Denis Kazakov, Robert V. Lindsey
Recurrent neural networks (RNNs) are difficult to train on sequence processing tasks, not only because input noise may be amplified through feedback, but also because any inaccurac…
Adapted Deep Embeddings: A Synthesis of Methods for -Shot Inductive Transfer Learning
Tyler R. Scott, Karl Ridgeway, Michael C. Mozer
The focus in machine learning has branched beyond training classifiers on a single task to investigating how previously acquired knowledge in a source domain can be leveraged to fa…
Learning Deep Disentangled Embeddings with the F-Statistic Loss
Karl Ridgeway, Michael C. Mozer
Deep-embedding methods aim to discover representations of a domain that make explicit the domain's class structure and thereby support few-shot learning. Disentangling methods aim…