I’m currently working towards a PhD in Machine Learning at University College London, under the supervision of David Barber. In 2020 and 2021 I interned at Google research, working on neural compression and incremental learning.

My general research interests include generative modelling, compression and representation learning.

Publications

• 3D Scene Compression through Entropy Penalized Neural Representation Functions[PDF]
  T Bird, J Ballé, S Singh, P Chou
  PCS 2021 (Learning-based Image Coding special session)
• Reducing the Computational Cost of Deep Generative Models with Binary Neural Networks[PDF]
  T Bird, F Kingma, D Barber
  ICLR 2021
• HiLLoC: lossless image compression with hierarchical latent variable models[PDF]
  J Townsend & T Bird (equal contribution), J Kunze, D Barber
  ICLR 2020
• Practical lossless compression with latent variables using bits back coding[PDF]
  J Townsend, T Bird, D Barber
  ICLR 2019
• Spread divergence[PDF]
  M Zhang, D Barber, T Bird, P Hayes, R Habib
  ICML 2020

Preprints

• Stochastic variational optimization [PDF]
  T Bird, J Kunze, D Barber
• Variational f-divergence minimization[PDF]
  M Zhang, T Bird, R Habib, T Zu, D Barber

More information on Google Scholar.

Random

Forest simulator

Contact

thomas dot bird at cs.ucl.ac.uk