Time |
Description (Presentation files available in Links below)
|
7:30-8:30 |
Registration – AAAI-20 |
8:30-8:35 |
Welcome and Introduction |
8:35-9:20 |
Keynote: Ece Kamar (Microsoft Research AI), AI in the Open World: Discovering Blind Spots of AI |
9:20-10:20 |
Session 1: Adversarial Machine Learning – Chair: Mauricio Castillo-Effen
– Bio-Inspired Adversarial Attack Against Deep Neural Networks, Bowei Xi, Yujie Chen, Fei Fan, Zhan Tu and Xinyan Deng.
– Nothing to See Here: Hiding Model Biases by Fooling Post hoc Explanation Methods, Dylan Slack, Sophie Hilgard, Emily Jia, Sameer Singh and Himabindu Lakkaraju.
– Adversarial Image Translation: Unrestricted Adversarial Examples in Face Recognition Systems, Kazuya Kakizaki and Kosuke Yoshida.
– Debate Panel – Paper Discussants: TBD
|
10:20-10:30 |
Poster Pitches 1 – (2 mins x pitch)
– Simple Continual Learning Strategies for Safer Classifers, Ashish Gaurav, Sachin Vernekar, Jaeyoung Lee, Vahdat Abdelzad, Krzysztof Czarnecki and Sean Sedwards.
– “How do I fool you?”: Manipulating User Trust via Misleading Black Box Explanations, Himabindu Lakkaraju and Osbert Bastani [not presented].
– Assessing the Adversarial Robustness of Monte Carlo and Distillation Methods for Deep Bayesian Neural Network Classification, Meet Vadera, Satya Narayan Shukla, Brian Jalaian and Benjamin Marlin.
– Fair Representation for Safe Artificial Intelligence via Adversarial Learning of Unbiased Information Bottleneck, Jin-Young Kim and Sung-Bae Cho.
– Out-of-Distribution Detection with Likelihoods Assigned by Deep Generative Models Using Multimodal Prior Distributions, Ryo Kamoi and Kei Kobayashi.
|
10:30-11:00 |
Poster Sessions and Coffee Break |
11:00-11:20 |
Invited Talk: François Terrier (Commissariat à l´Energie Atomique), Considerations for Evolutionary Qualification of Safety-Critical Systems with AI-based Components |
11:20-12:00 |
Session 2: Assurance Cases for AI-based Systems – Chair: Alec Banks
– Hazard Contribution Modes of Machine Learning Components, Ewen Denney, Ganesh Pai and Colin Smith.
– Assurance Argument Patterns and Processes for Machine Learning in Safety-Related Systems, Chiara Picardi, Colin Paterson, Richard Hawkins, Radu Calinescu and Ibrahim Habli.
– Debate Panel – Paper Discussants: Richard Mallah
|
12:00-12:10 |
Update Report: AI Safety Landscape Initiative, by Workshop Chairs |
12:10-12:50 |
Session 3: Considerations for the AI Safety Landscape – Chair: Huáscar Espinoza
– Founding The Domain of AI Forensics, Vahid Behzadan and Ibrahim Baggili.
– Exploring AI Safety in Degrees: Generality, Capability and Control, John Burden and José Hernández-Orallo.
– Debate Panel – Paper Discussants: Alec Banks
|
12:50-13:00 |
Poster Pitches 2 – (2 mins x pitch)
– SafeLife 1.0: Exploring Side Effects in Complex Environments, Carroll Wainwright and Peter Eckersley.
– (When) Is Truth-telling Favored in AI Debate?, Vojtech Kovarik and Ryan Carey.
– NewsBag: A Benchmark Multimodal Dataset for Fake News Detection, Sarthak Jindal, Raghav Sood, Richa Singh, Mayank Vatsa and Tanmoy Chakraborty.
– Algorithmic Discrimination: Formulation and Exploration in Deep Learning-based Face Biometrics, Ignacio Serna, Aythami Morales, Julian Fierrez, Manuel Cebrian, Nick Obradovich and Iyad Rahwan.
– Guiding Safe Reinforcement Learning Policies \\Using Structured Language Constraints, Bharat Prakash, Nicholas Waytowich, Ashwinkumar Ganesan, Tim Oates and Tinoosh Mohsenin.
|
13:00-14:00 |
Poster Sessions and Lunch (on your own; no sponsored lunch provided) |
14:00-14:20 |
Invited Talk: Sameer Singh (University of California, Irving), Evaluating and Testing Natural Language Processing Systems |
14:20-15:20 |
Session 4: Fairness and Bias – Chair: José Hernández-Orallo
– Fair Enough: Improving Fairness in Budget-Constrained Decision Making Using Confidence Thresholds, Michiel Bakker, Humberto Riveron Valdes, Duy Patrick Tu, Krishna Gummadi, Kush Varshney, Adrian Weller and Alex Pentland.
– A Study on Multimodal and Interactive Explanations for Visual Question Answering, Kamran Alipour, Jurgen P. Schulze, Yi Yao, Avi Ziskind and Giedrius Burachas.
– Models can be Learned to Conceal Unfairness from Explanation Methods, Botty Dimanov, Umang Bhatt, Mateja Jamnik and Adrian Weller.
– Debate Panel – Paper Discussants: Seán Ó hÉigeartaigh
|
15:20-15:30 |
Poster Pitches 3 – (2 mins x pitch)
– Practical Solutions for Machine Learning Safety in Autonomous Vehicles, Sina Mohseni, Mandar Pitale, Vasu Singh and Zhangyang Wang.
– Continuous Safe Learning Based on First Principles and Constraints for Autonomous Driving, Lifeng Liu, Yingxuan Zhu, Tim Tingqiu Yuan, and Jian Li.
– The Incentives that Shape Behavior, Ryan Carey, Eric Langlois, Tom Everitt and Shane Legg.
– Recurrent Neural Network Properties and their Verification with Monte Carlo Techniques, Dmitry Vengertsev and Elena Sherman.
– Toward Operational Safety Verification Via Hybrid Automata Mining Using I/O Traces of AI-Enabled CPS, Imane Lamrani, Ayan Banerjee and Sandeep Gupta.
|
15:30-16:00 |
Poster Sessions and Coffee Break |
16:00-17:20 |
Session 5: Uncertainty and Safe AI – Chair: Seán Ó hÉigeartaigh
– A Saddle-Point Dynamical System Approach for Robust Deep Learning, Yasaman Esfandiari, Keivan Ebrahimi, Aditya Balu, Umesh Vaidya, Nicola Elia and Soumik Sarkar.
– A High Probability Safety Guarantee with Shifted Neural Network Surrogates, Mélanie Ducoffe, Jayant Sen Gupta and Sebastien Gerchinovitz.
– Benchmarking Uncertainty Estimation Methods for Deep Learning With Safety-Related Metrics, Maximilian Henne, Adrian Schwaiger, Karsten Roscher and Gereon Weiss.
– PURSS: Towards Perceptual Uncertainty Aware Responsibility Sensitive Safety with ML, Rick Salay, Krzysztof Czarnecki, Maria Elli, Igancio Alvarez, Sean Sedwards and Jack Weast.
– Debate Panel – Paper Discussants: José Hernández-Orallo
|
17:20-17:30 |
Wrap-up and Best Paper Award |