Program

Download the full program in pdf here

All times are CET

15:00

Introduction  
Haralampos Stratigopoulos, Sorbonne Université, CNRS, LIP6
Ioana Vatajelu, Université Grenoble Alpes, CNRS, TIMA  
15:05












Invited presentations 1  
Session chair: Fei Su, Intel, USA  

15:05 Improving reliability of DNN accelerators
Kanad Basu, UT Dallas, USA  

15:30 Co-design of Quantized Neural Networks for Reliable Inference on FPGAs  
Giulio Gambardella, Xilinx, Ireland
 
15:55 Embracing the Unreliability of Memory Devices for Neuromorphic Computing
Damien Querlioz, Université Paris Saclay, France
 
16:20 Software Based Dependability Management of Neuromorphic Computing
Anup Das, Drexel University, USA
16:45Break
17:00









Invited presentations 2  
Session chair: Martin Andraud, Aalto University, Finland  

17:00 Robustness through Errors: Enhancing Machine Learning Security through Approximate Computing
Ihsen Alouani, Polytechnic University Hauts-De-France, France  

17:25 Reliability and Security for Machine Learning Systems
Muhammad Shafique, NYU Abu Dhabi, UAE, NYU, USA 

17:50 Functional Criticality Classification of Structural Faults in AI Accelerators (“ML for ML”)
Krish Chakrabarty, Duke University, USA
18:15Break
18:30







Panel: Test and dependability for AI chips: any different from traditional chips?  
Moderator: Mehdi Tahoori, KIT, Germany  

Panelists:
Alberto Bosio, INL, Ecole Centrale de Lyon, France
Yiorgos Makris, UT Dallas, USA
Siddharth Garg, NYU, USA
Eric Zhang, Horizon Robotics, Canada
Fadi Maamari, Synopsys, USA  
19:30

Closing remarks  
Haralampos Stratigopoulos, Sorbonne Université, CNRS, LIP6
Ioana Vatajelu, Université Grenoble Alpes, CNRS, TIMA