Publications

Conferences and PrePrints:

  1. Anirudh Goyal, Riashat Islam, DJ Strouse, Maxime Chevalier-Boisvert, Zafarali Ahmed, Doina Precup, Hugo Larochelle, Matthew Botvinick, Sergey Levine, Yoshua Bengio. Transfer and Exploration via the Information Bottleneck
    (under submission, ICLR 2019)
  2. Philip Bachman, Riashat Islam, Alessandro Sordoni, Zafarali Ahmed.
    “VFunc : A Deep Generative Model for Functions”. arxiv preprint, 2018 (Paper)
  3. Peter Henderson*, Riashat Islam*, Philip Bachman, Joelle Pineau, Doina Precup, David Meger. “Deep Reinforcement Learning that Matters”.  AAAI 2018 (Paper) (Science Magazine) (Blogpost)
  4. David Krueger*, Chin-Wei Huang*, Riashat Islam, Ryan Turner, Aaron Courville. “Bayesian Hypernetworks”.  Under review in UAI 2018 (Paper)
  5. Maziar Gomrokchi, Susan Amin, Riashat Islam, Doina Precup. “Persistence Length-Based Exploration in Deep Reinforcement Learning”. In Submission, 2017 
  6. Yarin Gal, Riashat Islam, Zoubin Ghahramani. “Deep Bayesian Active Learning with Image Data”. ICML 2017 (Paper | Code | Poster | Slides) 
  7. Bogdan Mazoure*, Riashat Islam*. “Alpha-Divergences in Variational Dropout” arxiv preprint, 2017 (Paper)
  8. H. M. Tarek Ullah, Zishan Ahmed Onik, Riashat Islam, Dip Nandi. “Alzheimer’s Disease And Dementia Detection From 3D Brain MRI Data Using Deep Convolutional Neural Networks.”. I2CT 2017

Workshops :

  1. Anirudh Goyal, Riashat Islam, Zafarali Ahmed, Doina Precup, Matthew Botvinick, Hugo Larochelle, Sergey Levine and Yoshua Bengio
    “InfoBot: Structured Exploration in Reinforcement Learning Using Information Bottleneck”. Exploration in Reinforcement Learning workshop, ICML 2018
  2. Philip Bachman, Riashat Islam, Zafarali Ahmed, Alessandro Sordoni.
    “VFunc : A Deep Generative Model for Functions”. Prediction and Generative Modelling in Reinforcement Learning workshop, ICML 2018 (Paper) (Oral)
  3. Khimya Khetarpal, Zafarali Ahmed, Andre Cianflone, Riashat Islam, Joelle Pineau “Re-Evaluate : Reproducibility in Evaluating Reinforcement Learning Algorithms”. Reproducibility in Machine Learning Workshop, ICML 2018 (Paper)
  4. Amy Zhang, Josh Romoff, Riashat Islam. “An Analysis of Optimization for Deep Reinforcement Learning”. Modern Trends in Nonconvex Optimization for Machine Learning workshop, ICML 2018
  5. David Krueger*, Chin-Wei Huang*, Riashat Islam, Ryan Turner, Alexandre Lacoste and Aaron Courville. “Bayesian Hypernetworks”.
    Bayesian Deep Learning workshop NIPS 2017 (Paper)
  6. Peter Henderson*, Thang Doan*, Riashat Islam and David Meger.
    “Bayesian Policy Gradients via Alpha-Divergence Dropout Inference”.
    Bayesian Deep Learning workshop, NIPS 2017 (Paper) (Poster)
  7. Riashat Islam*, Peter Henderson*, Maziar Gomrokchi, Doina Precup. “Reprodicibility of Benchmarked Deep Reinforcement Learning Tasks for Continuous Control”. Reproducibility in Machine Learning workshop, ICML 2017 (Paper | Poster | Code)
  8. Riashat Islam*, Peter Henderson*, Doina Precup.
    Contributed talk at Reinforcement Learning workshop, ICML 2017 (Slides)
  9. Maziar Gomrokchi, Susan Amin, Riashat Islam, Harsh Satija, Doina Precup. “Using Locally Self-Avoiding Random Walks for Exploration in Reinforcement Learning”. RLDM 2017
  10. Yarin Gal, Riashat Islam, Zoubin Ghahramani. “Active Learning with Image Data”. Bayesian Deep Learning workshop, NIPS 2016 (Paper | Poster)

* Equal Contributions

Journals :

  1. Vincent Francois-Lavet, Marc G. Bellemare, Peter Henderson, Riashat Islam, Joelle Pineau. “An Introduction to Deep Reinforcement Learning”.
    Under submission for Foundations and Trends in Machine Learning, 2018

Profile :


Google Scholar Profile

DBLP Profile

Arxiv

 

Conferences and Paper Reviewing :

  1. (NIPS 2017 Workshop) Bayesian Deep Learning
  2. NIPS 2018 Conference

 

Thesis :

  1. Riashat Islam, Yarin Gal, Zoubin Ghahramani. “Active Learning Image Data using Uncertainty in Deep Learning”. Masters Thesis, University of Cambridge, 2016 (Thesis | Poster | Code)
  2. Riashat Islam, Guy Lever, John Shawe-Taylor. “Improving Convergence of Deterministic Policy Gradient Methods in Reinforcement Learning”. Undergraduate Thesis, University College London, 2015 (Thesis | Code)

 

Others :

  1. Riashat Islam*, Peter Henderson*, Maziar Gomrokchi, Doina Precup. “Reproducibility of Benchmarked Deep Reinforcement Learning Tasks for Continuous Control”, Montreal AI Symposium, 2017


Other Notes (Unpublished) :

  1. Deterministic Intra-Option Policy Gradient Theorem (Report)
  2. Option Critic in Reproducing Kernel Hilbert Space (Report)

 

Technical Reports/Projects :

  1. Riashat Islam. “Exploration in Continuous Control Reinforcement Learning” (Slides)
  2. Riashat Islam, Catherine McGhan, Richard Murray. “Improved State Estimation and Control for Resilient Spacecraft Executive”.  SURF 2017, Caltech and NASA JPL (Report | Slides)
  3. Riashat Islam. “Large Vocabulary Speech Recognition” (Report)
  4. Riashat Islam. “Speech Recognition with GMM-HMMs” (Report)
  5. Riashat Islam, Jiameng Gao, Vera Johne. “Unifying Review of Variational Inference and Learning using Deep Directed Latent Variable Models” (Report | Poster)
  6. Riashat Islam. “Keyword Spotting” (Report)
  7. Riashat Islam. “Statistical Machine Translation” (Slides)
  8. Riashat Islam. “Reinforcement Learning for Spoken Dialogue Systems” (Report)
  9. Riashat Islam. “Basic Algorithms in Reinforcement Learning” (Report)
  10. Riashat Islam. “Playing Blackjack with Reinforcement Learning” (Report)
  11. Riashat Islam. “Cost-Sensitive Decision Tree Classifiers” (Report)
  12. Riashat Islam. “Gaussian Processes” (Report)
  13. Riashat Islam. “Weighted Automata using OpenFST” (Report)
  14. Riashat Islam. “Statistical Speech Synthesis” (Report)