Projects

Representation learning and disentangled representations

1 Disentangled Feature Representation for Few-shot Image Classification AAAI2022

2 Disentangling by Factorising ICML2018

Neuro-symbolic Learning and Reasoning

3 Generating Correct Answers for Progressive Matrices Intelligence Tests NeurIPS2020

4 Neural Abstract Reasoner NeurIPS2020

Continual Learning and Meta Reinforcement Learning

5 RL2- Fast Reinforcement Learning via Slow Reinforcement Learning ICLR2017

6 Learning to Continually Learn ECAI2020

7 Supermasks in Superposition NeurIPS2020

8 A Combinatorial Perspective on Transfer Learning NeurIPS2020

Recent papers in Meta-learning

9 Neural complexity measures NeurIPS2020

10 Adaptive Risk Minimization- Learning to Adapt to Domain Shift NeurIPS2021

11 Learning Novel Concepts by Imitating Drawings ICML2021

12 Meta-learning Deep Energy-based Memory Models ICLR2020

Recent papers in Neuro-symbolic Learning

13 Meta Module Network for Compositional Visual Reasoning NeurIPS2021

14 Neural Production Systems NeurIPS2021

15 Dynamic Inference with Neural Interpreters NeurIPS2021

Recent papers in Disentangled Representations

16 Visual Causal Feature Learning UAI2015

Recent Papers in Meta-reinforcement-learning

17 Disentangled Predictive Representation for Meta-Reinforcement Learning NeurIPS2021

18 A Consciousness-Inspired Agent for Reinforcement Learning NeurIPS2021

19 Markov State Abstractions for Deep Reinforcement Learning NeurIPS2021

Recent Papers in Continual learning

20 Continual Learning via Local Module Composition NeurIPS2021

21 DualNet Continual Learning Fast and Slow NeurIPS2021

22 Optimizing Reusable Knowledge for Continual Learning via Metalearning NeurIPS2021

Datasets for use in Projects

Visual - classification/continual-learning

shapes - shapes as used in class available here

Visual - classification/disentanglement/continual-learning

shapes in grid - shapes in a 3x3 grid available here

Time-series - classification

options feeds - option data available here and here

Tabular - classification

options - options data available here

market - market data available here

Tabular - reinforcement-learning

market-RL - market data in an RL setting available here

Tabular - continual-learning

market-CL - market data in a CL-setting available here

Visual - reasoning

few-shot NAR - data from the NAR paper in few-shot form; available here

Visual - reasoning

few-shot ARC - data from the ARC challenge from Chollet; available here

Code to load project datasets available here and here