Write a Blog >>
ICSE 2022
Sun 8 - Fri 27 May 2022

Reproducibility is an increasing concern in Artificial Intelligence (AI), particularly in the area of Deep Learning (DL). Being able to reproduce DL models is crucial for AI-based systems, as it is closely tied to various tasks like training, testing, debugging, and auditing. However, DL models are challenging to be reproduced due to issues like randomness in the software (e.g., DL algorithms) and non-determinism in the hardware (e.g., GPU). There are various practices to mitigate some of the aforementioned issues. However, many of them are either too intrusive or can only work for a specific usage context. In this paper, we propose a systematic approach to training reproducible DL models. Our approach includes three main parts: (1) a set of general criteria to thoroughly evaluate the reproducibility of DL models for two different domains, (2) a unified framework which leverages record-and-replay technique to mitigate software-related randomness and a profile-and-patch technique to control hardware-related non-determinism, and (3) a reproducibility guideline which explains the rationales and the mitigation strategies on conducting a reproducible training process for DL models. Case study results show our approach can successfully reproduce six open source and one commercial DL models.

Mon 9 May

Displayed time zone: Eastern Time (US & Canada) change

22:00 - 23:00
Machine Learning with and for SE 5Technical Track / Journal-First Papers / SEIP - Software Engineering in Practice at ICSE room 1-even hours
Chair(s): Jürgen Cito TU Wien and Meta
22:00
5m
Talk
Automatic Fault Detection for Deep Learning Programs Using Graph Transformations
Journal-First Papers
Amin Nikanjam École Polytechnique de Montréal, Houssem Ben Braiek École Polytechnique de Montréal, Mohammad Mehdi Morovati École Polytechnique de Montréal, Foutse Khomh Polytechnique Montréal
Link to publication DOI Media Attached
22:05
5m
Talk
Counterfactual Explanations for Models of Code
SEIP - Software Engineering in Practice
Jürgen Cito TU Wien and Meta, Isil Dillig University of Texas at Austin, Vijayaraghavan Murali Meta Platforms, Inc., Satish Chandra Facebook
Pre-print Media Attached
22:10
5m
Talk
VarCLR: Variable Semantic Representation Pre-training via Contrastive Learning
Technical Track
Qibin Chen Carnegie Mellon University, Jeremy Lacomis Carnegie Mellon University, Edward J. Schwartz Carnegie Mellon University Software Engineering Institute, Graham Neubig Carnegie Mellon University, Bogdan Vasilescu Carnegie Mellon University, USA, Claire Le Goues Carnegie Mellon University
DOI Pre-print Media Attached
22:15
5m
Talk
Towards Training Reproducible Deep Learning Models
Technical Track
Boyuan Chen Centre for Software Excellence, Huawei Canada, Mingzhi Wen Huawei Technologies, Yong Shi Huawei Technologies, Dayi Lin Centre for Software Excellence, Huawei, Canada, Gopi Krishnan Rajbahadur Centre for Software Excellence, Huawei, Canada, Zhen Ming (Jack) Jiang York University
Pre-print Media Attached
22:20
5m
Talk
Collaboration Challenges in Building ML-Enabled Systems: Communication, Documentation, Engineering, and ProcessDistinguished Paper Award
Technical Track
Nadia Nahar Carnegie Mellon University, Shurui Zhou University of Toronto, Grace Lewis Carnegie Mellon Software Engineering Institute, Christian Kästner Carnegie Mellon University
Pre-print Media Attached
22:25
5m
Talk
Detecting False Alarms from Automatic Static Analysis Tools: How Far are We?Nominated for Distinguished Paper
Technical Track
Hong Jin Kang Singapore Management University, Khai Loong Aw Singapore Management University, David Lo Singapore Management University
DOI Pre-print Media Attached

Wed 11 May

Displayed time zone: Eastern Time (US & Canada) change

11:00 - 12:00
Machine Learning with and for SE 10Technical Track / SEIP - Software Engineering in Practice / Journal-First Papers at ICSE room 1-odd hours
Chair(s): Preetha Chatterjee Drexel University, USA
11:00
5m
Talk
Defect Reduction Planning (using TimeLIME)
Journal-First Papers
Kewen Peng North Carolina State University, Tim Menzies North Carolina State University
Authorizer link Pre-print Media Attached
11:05
5m
Talk
Automatic Fault Detection for Deep Learning Programs Using Graph Transformations
Journal-First Papers
Amin Nikanjam École Polytechnique de Montréal, Houssem Ben Braiek École Polytechnique de Montréal, Mohammad Mehdi Morovati École Polytechnique de Montréal, Foutse Khomh Polytechnique Montréal
Link to publication DOI Media Attached
11:10
5m
Talk
Counterfactual Explanations for Models of Code
SEIP - Software Engineering in Practice
Jürgen Cito TU Wien and Meta, Isil Dillig University of Texas at Austin, Vijayaraghavan Murali Meta Platforms, Inc., Satish Chandra Facebook
Pre-print Media Attached
11:15
5m
Talk
VarCLR: Variable Semantic Representation Pre-training via Contrastive Learning
Technical Track
Qibin Chen Carnegie Mellon University, Jeremy Lacomis Carnegie Mellon University, Edward J. Schwartz Carnegie Mellon University Software Engineering Institute, Graham Neubig Carnegie Mellon University, Bogdan Vasilescu Carnegie Mellon University, USA, Claire Le Goues Carnegie Mellon University
DOI Pre-print Media Attached
11:20
5m
Talk
Towards Training Reproducible Deep Learning Models
Technical Track
Boyuan Chen Centre for Software Excellence, Huawei Canada, Mingzhi Wen Huawei Technologies, Yong Shi Huawei Technologies, Dayi Lin Centre for Software Excellence, Huawei, Canada, Gopi Krishnan Rajbahadur Centre for Software Excellence, Huawei, Canada, Zhen Ming (Jack) Jiang York University
Pre-print Media Attached
11:25
5m
Talk
Learning to Reduce False Positives in Analytic Bug Detectors
Technical Track
Anant Kharkar Microsoft, Roshanak Zilouchian Moghaddam Microsoft, Matthew Jin Microsoft Corporation, Xiaoyu Liu Microsoft Corporation, Xin Shi Microsoft Corporation, Colin Clement Microsoft, Neel Sundaresan Microsoft Corporation
Pre-print Media Attached
Hide past events

Information for Participants
Mon 9 May 2022 22:00 - 23:00 at ICSE room 1-even hours - Machine Learning with and for SE 5 Chair(s): Jürgen Cito
Info for room ICSE room 1-even hours:

Click here to go to the room on Midspace

Wed 11 May 2022 11:00 - 12:00 at ICSE room 1-odd hours - Machine Learning with and for SE 10 Chair(s): Preetha Chatterjee
Info for room ICSE room 1-odd hours:

Click here to go to the room on Midspace