2 code implementations • 16 Oct 2023 • Kieran Wood, Samuel Kessler, Stephen J. Roberts, Stefan Zohren
To deal with such situations, we propose a novel time-series trend-following forecaster that can quickly adapt to new market conditions, referred to as regimes.
1 code implementation • 4 Jan 2023 • Samuel Kessler, Adam Cobb, Tim G. J. Rudner, Stefan Zohren, Stephen J. Roberts
Sequential Bayesian inference can be used for continual learning to prevent catastrophic forgetting of past tasks and provide an informative prior when learning new tasks.
2 code implementations • 29 Nov 2022 • Samuel Kessler, Mateusz Ostaszewski, Michał Bortkiewicz, Mateusz Żarski, Maciej Wołczyk, Jack Parker-Holder, Stephen J. Roberts, Piotr Miłoś
World models power some of the most efficient reinforcement learning algorithms.
1 code implementation • 7 Feb 2022 • Bethan Thomas, Samuel Kessler, Salah Karout
In this paper we propose applying adapters to wav2vec 2. 0 to reduce the number of parameters required for downstream ASR tasks, and increase scalability of the model to multiple tasks or languages.
Automatic Speech Recognition Automatic Speech Recognition (ASR) +2
no code implementations • pproximateinference AABI Symposium 2022 • Samuel Kessler, Adam D. Cobb, Stefan Zohren, Stephen J. Roberts
Previous work in Continual Learning (CL) has used sequential Bayesian inference to prevent forgetting and accumulate knowledge from previous tasks.
no code implementations • 26 Jul 2021 • Samuel Kessler, Bethan Thomas, Salah Karout
We evaluate by applying these language representations to automatic speech recognition.
Automatic Speech Recognition Automatic Speech Recognition (ASR) +4
1 code implementation • 5 Jun 2021 • Samuel Kessler, Jack Parker-Holder, Philip Ball, Stefan Zohren, Stephen J. Roberts
In this paper we formalize this "interference" as distinct from the problem of forgetting.
no code implementations • 4 Dec 2019 • Samuel Kessler, Vu Nguyen, Stefan Zohren, Stephen Roberts
We place an Indian Buffet process (IBP) prior over the structure of a Bayesian Neural Network (BNN), thus allowing the complexity of the BNN to increase and decrease automatically.
1 code implementation • 8 Nov 2018 • Samuel Kessler, Arnold Salas, Vincent W. C. Tan, Stefan Zohren, Stephen Roberts
We introduce a novel framework for the estimation of the posterior distribution over the weights of a neural network, based on a new probabilistic interpretation of adaptive optimisation algorithms such as AdaGrad and Adam.