In the previous blog, we covered important points one must remember while preparing for a technical or research presentation. I have made the slides available in this Github repo containing some of the presentations I had personally prepared and presented in IIT Kharagpur.

1. Article : The spread of true and false news online, published in Science (March 2018 issue). In this presentation prepared by me and Amrith Krishna Da(a PhD scholar, CSE, IIT Kharagpur), we presented the above article. [PPT]

2. Topic : Semi-supervised Learning techniques and Active Learning [PPT]
I have only provided my segment, which was a part of a panel discussion covering a broader topic titled Leveraging Unlabeled Data and Environment Access for ML. In the discussion panel, we also covered recent literature in Transfer learning, Zero-shot learning, Reinforcement Learning(with different variants) and finally, Imitation Learning.

The following papers were discussed :

Semi-supervised learning :

  1. Active Learning for Convolutional Neural Networks, ICLR 2018
  2. Estimating Accuracy from Unlabelled Data, NIPS 2017
  3. When does label propagation fail? a view from a network generative model, IJCAI ‘17
  4. Cost-effective training of deep cnns with active model adaptation, KDD 2018

Reinforcement learning

  1. FFNet: Video Fast Forwarding via Reinforcement Learning, CVPR 2018
  2. Human-level control through deep reinforcement learning, Nature 2015

Transfer, multi-task and few shot learning

a) One Shot Imitation Learning , NIPS 2017

b) When will You Arrive, Estimating Travel Time Based on Deep Neural Networks, AAAI 2018 (Multi task Learning)

      1. Transfer Learning
a) Universal Language Model Fine-Tuning for Text Classification (ULMFit) ACL, 2018

 b) Deep contextualized word representations (ELMo) NAACL, 2018 ]

  1. Multi-task Learning

A Hierarchical Multi-task Approach for Learning Embeddings from Semantic Tasks AAAI, 2019

  1. Few shot Learning

a) High-risk learning: acquiring new word vectors from tiny data  EMNLP 2017 (short paper)

b) Zero-shot Learning of Classifiers from Natural Language Quantification ACL 2018