Felix Leibfried

Senior Quantitative Researcher

Eisler Capital, Cambridge, UK

Contact me at felix.leibfried@gmail.com.

Since May 2021, I am a senior quantitative researcher at London-based hedge fund Eisler Capital in their Cambridge office, working on systematic trading strategies.


Prior to that, I was a machine learning research team lead at Secondmind (formerly PROWLER.io), a Cambridge-based start-up for autonomous decision-making in industrial applications. My team and I developed Bellman, an open-source software library for model-based reinforcement learning with the aim to support basic research and customer-related applications. We were also partially involved in another open-sourcing effort leading to GPflux, a library for Bayesian deep learning covering Bayesian neural networks and deep Gaussian processes to provide flexible probabilistic models for approximate inference.

About


In the past, my basic research at  Secondmind mostly focused on reinforcement learning with specific emphasis on information-theoretically motivated algorithms, but I was also involved in research around Bayesian deep learning. On the applied side, I was involved in a couple of customer-related projects, e.g. developing a model-based reinforcement learning agent for systematic trading, creating a framework for tuneable artificial intelligence, and project-leading efforts to develop customised Gaussian-process solutions for recommendation systems


Before Secondmind, I interned at Microsoft Research in Cambridge developing predictive models for Atari-style video games. The specific problem that we solved was to jointly model image sequence and reward prediction in such games (which had been unresolved until then).

I completed my PhD in artificial intelligence at the Max Planck Institutes for Intelligent Systems and Biological Cybernetics in Tuebingen, Germany. My educational background is computer science with focus on artificial intelligence (eq. BSc+MSc) but I also obtained a degree in biology with focus on neuroscience (eq. BSc+MSc), both from the University of Wuerzburg, Germany.

Publications (see also Google Scholar)


2022 Leibfried F

Variational inference for model-free and model-based reinforcement learning

arXiv


2021 Dutordoir V, Salimbeni H, Hambro E, McLeod J, Leibfried F, Artemev A, van der Wilk M, Hensman J, Deisenroth MP and John ST

GPflux: A library for deep Gaussian processes

arXiv


2021 McLeod J, Stojic H, Adam V, Kim D, Grau-Moya J, Vrancx P and Leibfried F

Bellman: A toolbox for model-based reinforcement learning in TensorFlow

arXiv


2020 Leibfried F, Dutordoir V, John ST and Durrande N

A tutorial on sparse Gaussian processes and variational inference

arXiv


2020 Pearce T, Leibfried F, Brintrup A, Zaki M and Neely A

Uncertainty in neural networks: Approximately Bayesian ensembling

Proceedings of the Conference on Artificial Intelligence and Statistics (AISTATS)


2019 Leibfried F, Pascual-Diaz S and Grau-Moya J

A unified Bellman optimality principle combining reward maximization and empowerment

Proceedings of the Conference on Neural Information Processing Systems (NeurIPS)


2019 Leibfried F and Grau-Moya J

Mutual-information regularization in Markov decision processes and actor-critic learning

Proceedings of the Conference on Robot Learning (CoRL)


2019 Grau-Moya J, Leibfried F and Vrancx P

Soft Q-learning with mutual-information regularization

Proceedings of the International Conference on Learning Representations (ICLR)


2018 Leibfried F and Vrancx P

Model-based regularization for deep reinforcement learning with transcoder networks

Workshop on Deep Reinforcement Learning at the Conference on Neural Information Processing Systems (NeurIPS Workshop)


2018 Leibfried F, Grau-Moya J and Bou-Ammar H

An information-theoretic optimality principle for deep reinforcement learning

Workshop on Deep Reinforcement Learning at the Conference on Neural Information Processing Systems (NeurIPS Workshop)


2018 Grau-Moya J, Leibfried F and Bou-Ammar H

Balancing two-player stochastic games with soft Q-learning

Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI)


2017 Leibfried F, Kushman N and Hofmann K

A deep learning approach for joint video frame and reward prediction in Atari games

Workshop on Principled Approaches to Deep Learning at the International Conference on Machine Learning (ICML Workshop)


2017 Peng Z, Genewein T, Leibfried F and Braun DA

An information-theoretic on-line update principle for perception-action coupling

Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)


2016 Grau-Moya J, Leibfried F, Genewein T and Braun DA

Planning with information-processing constraints and model uncertainty in Markov decision processes

Proceedings of the European Conference on Machine Learning (ECML PKDD)


2016 Leibfried F and Braun DA

Bounded rational decision-making in feedforward neural networks

Proceedings of the Conference on Uncertainty in Artificial Intelligence (UAI)


2015 Genewein T, Leibfried F, Grau-Moya J and Braun DA

Bounded rationality, abstraction, and hierarchical decision-making: An information-theoretic optimality principle

Frontiers in Robotics and AI


2015 Leibfried F and Braun DA

A reward-maximizing spiking neuron as a bounded rational decision maker

Neural Computation


2015 Leibfried F, Grau-Moya J and Braun DA

Signaling equilibria in sensorimotor interactions

Cognition

Patent Applications


2017 Kim D, Nicholson T, Ferguson N, ..., Hensman J, Leibfried F, Grau-Moya J, John ST, Bou-Ammar H and Tukiainen A

System architecture for an artificial intelligence platform


2017 Grau-Moya J, Leibfried F and Bou-Ammar H

Tuneable artificial intelligence