9:00 - 9:25: Autonomous Model Management via Reinforcement Learning, Elad Liebman, Eric Zavesky and Peter Stone.
9:25 - 9:50 Optimizing Hierarchical Classification with Adaptive Node Collapses, Sujan Perera, Orna Raz, Sheng Hua Bao, Ramani Routray and Marcel Zalmanovici.
9:50 - 10:15 Can We Achieve Open Category Detection with Guarantees? Si Liu, Risheek Garrepalli, Alan Fern and Tom Dietterich. presentation
10:15 - 10:40 Distributed Deep Learning under Differential Privacy with the Teacher-Student Paradigm, Jun Zhao.
11:00 - 11: 25: Denoising Dictionary Learning Against Adversarial Perturbations, John Mitro. presentation
11:25 - 11:35: Determining classification performance degradation by analyzing changes in the data distribution, Ilan Shimshoni.
11:35-12:00 MTDeep: Boosting the Security of Deep Neural Nets Against Adversarial Attacks with Moving Target Defense, Sailik Sengupta, Tathagata Chakraborti and Subbarao Kambhampati.
12:00 - 12:10 Defending via strategic ML selection, Eitan Farchi and Onn Shehory.
12:10 - 12:20 Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks, Sihao Ding and Andreas Wallin.
14:00 - 14:40 Invited talk: Using Combinatorial Test Design to Test An Automated Driving Car (see abstract below) , Satoshi Masuda , IBM Research, Tokyo, Japan.
14:40 - 15:05 Multiple-Implementation Testing of Supervised Learning Software, Siwakorn Srisakaokul, Zhengkai Wu, Angello Astorga, Oreoluwa Alebiosu and Tao Xie. presentation
15:05 - 15:15 Quality Assurance Framework for Artificial Intelligence, Yasuharu Nishi, Satoshi Masuda, Hideto Ogawa and Keiji Uetsuki.
15:15 - 15:25 An Investigation of Bounded Misclassification for Operational Security of Deep Neural Networks, Sailik Sengupta, Andrew Dudley, Tathagata Chakraborti, Subbarao Kambhampati.
16:00 - 16:25: Climbing the Kaggle Leaderboard by Exploiting the Log-Loss Oracle, Jacob Whitehill. presentation
16:25 - 16:50: Telemade: A Testing Framework for Learning-Based Malware Detection Systems, Wei Yang and Tao Xie. presentation
16:50 - 17:15 Evaluation of Predictive Models for Wildlife Poaching Prediction through Controlled Field Test in Uganda, Shahrzad Gholami, Benjamin Ford, Debarun Kar, Fei Fang, Andrew Plumptre, Milind Tambe, Margaret Driciru, Fred Wanyama, Aggrey Rwetsiba, Mustapha Nsubaga, Joshua Mabonga and Bistra Dilkina. presentation
17:15 - 17:30: Summary and wrapping up
Abstract of the invited talk "Using Combinatorial Test Design to Test An Automated Driving Car "
Research and development in the field of automated vehicles has increased along with related work on automated driving (AD) software. Thorough testing of AD software using simulation must be conducted in advance of testing AD cars on the road. Parameters of the many objects around an AD car, such as other cars, traffic lanes, and pedestrians, are required as inputs of the simulation. Therefore, the number of parameter combinations becomes extremely large. A combination of parameters is called as a test case; hence, the challenge is to reduce the number of test cases. In this talk, I present an industry case study about using combinatorial test design (CTD) to test an automated driving car. The talk includes, overview of testing an AD car, how CTD itself works, a problem of applying CTD to testing an AD car and approaches to solve the problem. The problem is that CTD reduces combinatorial test cases based on empirical data of software faults originally, and the data did not base on objects around an AD car. We analyzed and experimented data optimization for applying CTD to testing AD car on high-way three lane.
Dr. Satoshi Masuda is a Senior Researcher at Security & Services , IBM Research - Tokyo.