Machine Learning Reading Group

ML group is a machine learning reading group at Purdue ECE, coordinated by Prof Stanley Chan. The purpose of the group is to provide a platform for intellectual discussions on latest topics in machine learning.

The reading group consists of a weekly meeting and a mailing list. We welcome all faculty and students to participate.

Mailing List & Piazza Forum

Please subscribe our mailing list at

https://engineering.purdue.edu/ECN/mailman/listinfo/machine-learning

If you like to participate in our Piazza forum, you may sign up at

https://piazza.com/purdue/other/ece00000

Schedule (Spring 2020)

Copyrights of the slides are owned by individual faculty and students, or sources where the slides are originally from. Please contact the presenter if you have any question regarding the material.

Time: Every Tuesday 2-3pm
Location: MSEE 239
What: Faculty and student presenting papers

Spring 2020

The following are 4 categories of papers we are going to study. If you are interested in presenting one or more of the papers, please let me know.

Topic 1: Modular Design Framework
Paper 1.1: (taken) Consensus Equilibrium
Paper 1.2: Modular Representation of Layered Neural Networks
Paper 1.3: Modular Networks: Learning to Decompose Neural Computation
Paper 1.4: (taken) Deep Density Destructors
Paper 1.5: (taken) Putting an End to End-to-End: Gradient-Isolated Learning of Representations

Topic 2: Signal Processing for Deep Learning
Paper 2.1: The Geometry of Deep Networks: Power Diagram Subdivision
Paper 2.2: (taken) Deep Convolutional Framelets: A General Deep Learning Framework for Inverse Problem
Paper 2.3: (taken) Mad Max: Affine Spline Insights into Deep Learning
Paper 2.4: (taken) Interpretable Convolutional Neural Networks via Feedforward Design

Topic 3: Interpretable AI
Paper 3.1: This Looks Like That: Deep Learning for Interpretable Image Recognition
Paper 3.2: Learning to Explain With Complemental Examples
Paper 3.3: Learning to Explain: An Information-Theoretic Perspective on Model Interpretation
Paper 3.4: GNNExplainer: Generating Explanations for Graph Neural Networks
Paper 3.5: Towards Automatic Concept-based Explanations
Paper 3.6: INVASE: Instance-wise Variable Selection using Neural Networks
Paper 3.7: (taken) A Unified Approach to Interpreting Model Predictions

Topic 4: Recent Advances in Deep Learning
Paper 4.1 (taken) Reconciling modern machine learning practice and the bias-variance trade-off
Paper 4.2 Critical Learning Periods in Deep Neural Networks
Paper 4.3 (taken) Robustness May Be at Odds with Accuracy
Paper 4.4 (taken) Understanding GANs: the LQG Setting
Paper 4.5 Towards understanding knowledge distillation
Paper 4.6 When Does SGD Escape Local Minima?
Paper 4.7 (taken) How Does Batch Normalization Help Optimization?

April 28, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper: Justin Yang Paper 2.4 Interpretable Convolutional Neural Networks via Feedforward Design

April 21, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper: Kent Gauen Paper 2.3 Mad Max: Affine Spline Insights into Deep Learning

April 14, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper: Ankit Manerikar Deep Convolutional Framelets: A General Deep Learning Framework for Inverse Problem

April 7, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Faculty Meeting. Not for Public.

March 31, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper: Fangda Li Paper 4.7 How Does Batch Normalization Help Optimization?

March 24, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper 1: Rui Wang Paper 3.7 A Unified Approach to Interpreting Model Predictions
Paper 2:

March 17, 2020 (Tuesday)
Spring Break

March 10, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper 1: Yash Sanghvi Paper 4.1 Reconciling modern machine learning practice and the bias-variance trade-off
Paper 2:

March 3, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Faculty Meeting. Not for Public.

February 25, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper 1: Stanley Chan Paper 1.1 An overview of Plug-and-Play Priors
Paper 2: Diyu Yang Paper 1.1 Consensus Equilibrium

February 18, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper 1: Madhuri Nagare Paper 4.4 Understanding GANs: the LQG Setting

February 11, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Presenter: Prof. David Inouye
Paper 1: Paper 1.4 Deep Density Destructors
Paper 2: Paper 1.5 Putting an End to End-to-End: Gradient-Isolated Learning of Representations

February 4, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Faculty Meeting. Not for Public.

January 28, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Paper: Stanley Chan Paper 4.3 Robustness May Be at Odds with Accuracy

January 21, 2020 (Tuesday)
Time: 2pm-3pm
Room: MSEE 239
Meeting Canceled.

Fall 2019

December 4, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: S. Shams, Deep Mutual Learning
Paper 2: Zhiyuan Mao, Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors

November 27, 2019
Thanksgiving.

November 20, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Henry Wang, Pattern classification with missing data: a review
Paper 2: Rehana Mahfaz, BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

November 13, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Chang-Shen Lee, Collaborative Deep Learning for Recommender Systems
Paper 2: Yinghan Long, Non-local neural networks

November 6, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Junghoon Kim, Inference and Missing Data
Paper 2: Ankit Manerikar, Speed versus accuracy in visual search: Optimal performance and neural architecture

October 30, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Tony Allen, Geometric deep learning on graphs and manifolds using mixture model CNNs
Paper 2: Yash Sanghvi, Seeing into Darkness: Scotopic Visual Recognition

October 23, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Nick Chimitt, The Unreasonable Effectiveness of Deep Features as a Perceptual Metric
Paper 2: Madhuri Nagare, Image-to-Image Translation with Conditional Adversarial Nets

October 16, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper: Prof. Chris Brinton
Title: TBD

October 9, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Kent Gauen, Convolutional Sequence to Sequence Learning
Paper 2: Fangda Li, Large scale GAN training for high fidelity natural image synthesis

October 2, 2019
No meeting

September 25, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Speaker: Prof. Joy Wang
Title: Learning from Limited Labeled Data via Generative Adversarial Network

September 18, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Di Chen, Every Moment Counts: Dense Detailed Labeling of Actions in Complex Videos
Paper 2: Isha Garg, Opening the black box of Deep Neural Networks via Information

September 11, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Yue Cao, Deep Reinforcement Learning for Robotic Manipulation
Paper 2: Maliha Hossain, Loss Functions for Image Restoration with Neural Networks

September 4, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Speaker: Prof. David Inouye
Title: Deeper Understanding via Destructive Deep Learning and Counterfactual Explanations

August 28, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Intro: A 10-minute Introduction to ML Reading Group. Tips on Presentation by Prof. Buzzard
Talk 1: Stanley Chan, Detecting Photoshopped Faces by Scripting Photoshop
Talk 2: Abhiram Gnanasambandam, Dirty Pixels: Optimizing Image Classification Architectures for Raw Sensor Data

Summer 2019

August 7, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Abdu Alanazi, Deep Image Prior
Paper 2: canceled

July 31, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Tony Allen, Building Machines That Learn and Think Like People
Paper 2: Yiheng Chi, Theoretically Principled Trade-off between Robustness and Accuracy

July 24, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Hao Li, Neural Ordinary Differential Equations
Paper 2: Grant Bowman, A Neural Algorithm of Artistic Style

July 17, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Qingshuang Chen, Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation
Paper 2: TBD, Tracking the World State with Recurrent Entity Networks

July 10, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Fernando Davis, Analysis of classifiers’ robustness to adversarial perturbations
Paper 2: He Liu, Improving drug discovery with high-content phenotypic screens by systematic selection of reporter cell lines

July 3, 2019
TBD

Jun 26, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Xiwen Zhang, Exploring Randomly Wired Neural Networks for Image Recognition
Paper 2: Haoyu Chen, Who Said What: Modeling Individual Labelers Improves Classification

Jun 19, 2019
Time: 2:30-3:30pm
Room: MSEE 239
Paper 1: Nick Chimitt, Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
Paper 2: Zhiyuan Mao, Deep learning applications for predicting pharmacological properties of drugs and drug repurposing using transcriptomic data

Spring 2019

Apr 24, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Kent Gauen, Understanding deep learning requires rethinking generalization
Paper 2: MeherChaitanya Pindiprolu and Pal Raktim, An Actor-Critic Algorithm for Sequence Prediction

Apr 17, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Beheshteh Rakhshan and Shixin Zheng, Explaining and harnessing adversarial examples
Paper 2: Ryan Dailey and Mridul Agarwal, Stronger generalization bounds for deep nets via a compression approach

Apr 10, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Dongqi Zheng and Amrit Nagarajan, Understanding deep convolutional networks
Paper 2: Xiaodong Huang and Yubo Wang, How SGD Selects the Global Minima in Over-parameterized Learning: A Dynamical Stability Perspective

Apr 3, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Christopher Wright and Jacob Desmond, In Ictu Oculi: Exposing AI Generated Fake Face Videos by Detecting Eye Blinking
Paper 2: Brian Page and Liyao Gao, Learning to Optimize Neural Nets

Mar 27, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Alex Carignan and Zachary Letterhos, Auto-Encoding Variational Bayes
Paper 2: Tony Allen and Gavin Glenn, Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm

Mar 20, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Guanzhe Hong, Quantum machine learning
Paper 2: Yifan Wang and Yang Zhang, Improved Techniques for Training GANs

Mar 13, 2019
Spring Break

Mar 6, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Natat Sombuntham, Obfuscated Gradients Give a False Sense of Security
Paper 2: Omar Eldaghar and Alden Bradford, Audio Adversarial Examples: Targeted Attacks on Speech-to-Text

Feb 27, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Rehana Mahfuz, Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
Paper 2: Yizhen Zhang and Yujie Chen, Attention Is All You Need

Feb 20, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Omar Elgendy, Image Super-Resolution Using Deep Convolutional Networks
Paper 2: Sthitapragyan Parida and Ziyu Gong, Empirical study of the topology and geometry of deep networks

Feb 13, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Ye Shi and Shiqi Zhang, Discriminative vs Informative Learning
Paper 2: Carlos Salinas and Max Ruby, Approximation by Superpositions of a Sigmoidal Function

Feb 6, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Jingshuang Chen and Xiaofeng Ou, Learning Functions: When Is Deep Better than Shallow?
Paper 2: Abhiram Gnanasambandam and Vineeth Ravi, Deep Convolutional Framelets: A General Deep Learning Framework for Inverse Problems

Jan 30, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Xiangyu Qu, Mask R-CNN
Paper 2: Lakshay Kharbanda and Rahul Nanda, Machine Learning: The High-Interest Credit Card of Technical Debt

Jan 23, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Omkar Patil and Parameswaran Desigavinayagam, U-Net: Convolutional Networks for Biomedical Image Segmentation
Paper 2: Maliha Hossain and Balke Thilo, Deep Residual Learning for Image Recognition

Jan 16, 2019
Time: 2:30-3:30pm
Room: WTHR 320
Paper 1: Diyu Yang, A Survey of Model Compression and Acceleration for Deep Neural Networks
Paper 2: Omar Elgendy, A Review of Convolutional Neural Networks for Inverse Problems in Imaging

Jan 9, 2019
No Meeting

Fall 2018

Dec 5, 2018
No Reading Group. Final Exam

Nov 28, 2018
Time: 2:30-3:30pm
Room: MSEE 239
Title: TBD
Presenster: Jing Li
Reading 1:

Nov 21, 2018
Title: Thanksgiving Break

Nov 14, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: Perceptual Learning
Presenster: Prof Amy Reibman
Reading 1:

Nov 7, 2018
Time: 2:30-3:30pm
Room: MSEE 239
Title: Graph Convolutional Networks
Presenster: David Ho
Reading 1: https://arxiv.org/abs/1609.02907

Oct 31, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: Doing Deep Learning on ITaP Clusters
Presenster: Amiya Maji
Reading 1: Overview of RCAC

Oct 24, 2018
Time: 2:30-3:30pm
Room: MSEE 239
Title: Empirical Study of the Topology and Geometry of Deep Networks
Presenster: Xiangyu Qu
Reading 1: Fawzi et al. CVPR 2018

Oct 17, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: Wasserstein GAN
Presenster: Omar Elgendy
Reading 1: https://arxiv.org/abs/1701.07875

Oct 10, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: Linear Factor Analysis and Auto-encoder
Presenster: Rajeev Sahay
Reading 1: Deep Learning Book Ch. 13 and Ch. 14

Oct 3, 2018
Room: MSEE 239
Title: Evaluating the Robustness of Neural Networks: An Extreme Value Theory Approach
Presenster: Guanzhe Hong
Reading 1: https://arxiv.org/abs/1801.10578

Sep 26, 2018
Room: POTR 234 Fu Room
Title: Neural Network for Universal Approximation
Presenster: Sri Kalyan Yarlagadda
Reading 1: Multilayer Feedforward Networks are Universal Approximators

Sep 19, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: CNSIP Seminar: Signal and Information Processing in the Age of Massive Data: Exploiting the Blessings of Dimensionality
Presenster: Prof Yue M. Lu, Harvard John Paulson School of Engineering and Applied Sciences

Sep 12, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: Image-Based 3D Scene Understanding: From Camera Geometry to Deep Learning
Presenster: Shaobo Fang

Sep 5, 2018
Time: 2:30-3:30pm
Room: MSEE 239
Title: Practical Considerations of Deep Networks
Presenster: Diyu Yang
Reading 1: Deep Learning Textbook Ch.11

Aug 29, 2018
Time: 2:30-3:30pm
Room: POTR 234 Fu Room
Title: Deep Learning Research in Video and Image Processing Lab
Presenster: Prof. Edward Delp
Reading 1: VIPER website

Summer 2018

Aug 08, 2018
Deep Learning Summer Workshop 2018 Part 2

Aug 01, 2018
Title: Reinforcement Learning
Presenster: Prof. Charles Bouman
Reading 1: Useful links

July 25, 2018
Title: Adversarial Attack
Presenter: Prof. Stanley Chan
Reading 1: Explaining and Harnessing Adversarial Examples’’, by Goodfellow et al.

July 18, 2018
Title: Recurrent and Recursive Networks
Presenter: Prof. Aly El Gamal
Reading 1: Deep Learning Textbook, Chapter 10

July 11, 2018
Title: Feed Forward Network and Back Propagation
Presenter: Soumendu Majee

Reading 1: Deep Learning Textbook, Chapter 6

July 04, 2018
No Meeting.
Independence Day

June 27, 2018
Title: Convolutional Neural Networks
Presenter: Prof. Maggie Zhu
Reading 1: Stanford CS 231 Note
Reading 2: Deep Learning Textbook, Chapter 9

June 20, 2018
Deep Learning Summer Workshop 2018 Part 1

June 13, 2018
Title: TensorFlow Introduction
Presenter: Amir Ziabari
Reading 1: TensorFlow Tutorial

June 06, 2018
Title: PyTorch Introduction
Presenter: Xiangyu Qu
Reading 1: PyTorch Tutorial

May 30, 2018
Title: Deep Learning Optimization
Presenter: Prof. Aly El Gamal
Reading 1: Deep Learning Textbook Chapter 8

May 23, 2018
Title: Generative Adversarial Networks
Presenter: Prof. Stanley Chan
Reading 1: ‘‘NIPS 2016 Tutorial: Generative Adversarial Networks’’ by Goodfellow
Reading 2: ‘‘Generative Adversarial Nets’’ by Goodfellow et al.

May 16, 2018
Title: Overview of Deep Learning
Presenter: Prof. Greg Buzzard
Reading 1: Deep Learning Textbook Chapter 1