Dharmesh Tailor

I am a research assistant in the Approximate Bayesian Inference Team at the RIKEN Centre for Advanced Intelligence Project (Tokyo, Japan).

From 2017 to 2018, I worked at the European Space Agency as a ‘Young Graduate Trainee’ in the Advanced Concepts Team (Netherlands). I worked with Dr. Dario Izzo on a policy search method combining imitation learning and trajectory optimization.

In 2017, I graduated with an MSc in Artificial Intelligence from the University of Edinburgh specialising in machine learning and computational neuroscience. I did my thesis under Prof. Mark van Rossum on (neuronal) population coding. I also worked as a teaching assistant for IAML (computer lab tutorials) and DMMR (small-group teaching).

In 2016, I graduated with a Bachelors in Computer Science and Mathematics (joint course) from Imperial College London. During the summers, I did software internships at Ocado Technology and Diamond Light Source.

Email  /  GitHub  /  Google Scholar

profile photo

Highlights / News

March 26, 2021: Will be joining the Amsterdam Machine Learning Lab as a PhD candidate supervised by Eric Nalisnick in September

March 10, 2021: Talk on Identifying Memorable Experiences of Learning Machines at RIKEN AIP Open Seminar


I’m currently interested in variational inference, kernel methods (including Gaussian processes) and reinforcement learning. Previously I have worked on topics in optimal control/trajectory optimization and computational neuroscience.

project image

On the Stability Analysis of Deep Neural Network Representations of an Optimal State-Feedback

Dario Izzo, Dharmesh Tailor, Thomas Vasileiou
IEEE Transactions on Aerospace and Electronic Systems, 2020
paper / arxiv /

Developed a novel method to perform high-order analysis of motion stability for a neurocontrolled system using differential algebraic techniques.

project image

Learning the Optimal State-Feedback via Supervised Imitation Learning

Dharmesh Tailor, Dario Izzo
Astrodynamics (Springer), 2019
paper / arxiv / code /

A deep neural network is trained to approximate the optimal state-feedback in a deterministic and continuous-time setting from solutions of a trajectory optimiser. The approach is demonstrated for a quadcopter model with quadratic and time-optimal objective functions.

project image

Machine Learning and Evolutionary Techniques in Interplanetary Trajectory Design

Dario Izzo, Christopher Sprague, Dharmesh Tailor
Modeling and Optimization in Space Engineering (Springer), 2019
paper / arxiv /

A deep neural network is trained to represent the optimal guidance profile of an Earth–Mars orbital transfer.

project image

Unconscious Biases in Neural Populations Coding Multiple Stimuli

Sander Keemink, Dharmesh Tailor, Mark van Rossum
Neural Computation, 2018
paper /

When multiple stimuli that overlap in their neural representation are simultaneously encoded in the population, biases in the read-out emerge. To study the origin of the bias, a framework based on Gaussian Processes is developed that allows an accurate calculation of the estimate distributions of maximum likelihood decoders.

Other notes

Adapted from Leonid Keselman's fork of John Barron's website.