25 citations · 44 across the 4 of their papers we have counts for
4 papers
Hierarchical Quantized Autoencoders
Will Williams, Sam Ringer, Tom Ash +3
Despite progress in training neural networks for lossy image compression, current approaches fail to maintain both perceptual quality and abstract features at very low bitrates. En…
Texture Bias Of CNNs Limits Few-Shot Classification Performance
Sam Ringer, Will Williams, Tom Ash +2
Accurate image classification given small amounts of labelled data (few-shot classification) remains an open problem in computer vision. In this work we examine how the known textu…
Discriminative training of RNNLMs with the average word error criterion
Rémi Francis, Tom Ash, Will Williams
In automatic speech recognition (ASR), recurrent neural language models (RNNLM) are typically used to refine hypotheses in the form of lattices or n-best lists, which are generated…
Scaling Recurrent Neural Network Language Models
Will Williams, Niranjani Prasad, David Mrva +2
This paper investigates the scaling properties of Recurrent Neural Network Language Models (RNNLMs). We discuss how to train very large RNNs on GPUs and address the questions of ho…