Industry's panel this morning at #SEMLA2019

All great things have to come to an end. Thanks to all the OC of #icse19 & a huge thanks to our wonderful general chair Jo Atlee

Happy to introduce a new task and benchmark for lifelong learning: "IIRC: Incremental Implicitly-Refined Classification". arxiv.org/abs/2012.12477 Website: chandar-lab.github.io/IIRC/. Work led by my student Mohamed in collaboration with Mojtaba, Shagun Sodhani, and myself. 1/n

Hello World! We are finally on Twitter! Chandar Lab is a research group based in Polytechnique Mtl and Mila - Institut québécois d'IA led by Sarath Chandar. Our work focuses on #LifelongLearning, #DeepLearning, #ReinforcementLearning, and #NLProc. Follow us for the latest news on our work!

Recent discussions suggests that #optimization algorithms have reached their full potential. In our new work, we explore the robust notion of Memory Augmentation and the improvements it offers to commonly used optimizers. Work led by Paul-Aymeric McRae and Prasanna.

It's happening now. The YouTube streaming link for those who missed to be here today at #SEMLA2022: youtu.be/QjS7FdHfzvA

A Talk by Prof. Ipek Ozkaya on "Are you ready to engineer and sustain AI systems?". Keynote: "We are bridging software and AI. If we don't bridge them well, we will cause confusion and end up with low-quality software". YouTube streaming link: youtu.be/QjS7FdHfzvA #SEMLA2022

A great talk by Prof. Qinghua Lu on "Operationalising Responsible AI at Scale". She talked about the challenges and solutions beyond algorithms for #responsibleAI. Keynote: "We should turn principles into practice". YouTube Streaming: youtu.be/QjS7FdHfzvA #SEMLA2022

Prof. Reihaneh Rabbany from McGill University gave an amazing talk on "Analyzing Complex Data from Online Societies" and presented how social media and politics intersect by highlighting the impact of fake news and rumors on social media. YouTube streaming: youtu.be/QjS7FdHfzvA #SEMLA2022

The keynotes of our first-panel session #SEMLA2022: + We should #integrate our solutions for #responsibleAI in production regarding deployment and cost. + Making progress on #generalization is a critical issue. + We have a lot of #guidance and now, we need tools to apply them.

Professor Hridesh Rajan from Iowa State University gave a great talk on "More Modular Deep Learning", presented outstanding results on: + Can we fix bugs in (or add features to) trained models by #decomposing into #modules? + What is the performance like in #deeplearning models? #SEMLA2022

#Poster lightning talks are now being showcased on #CSERspr23. Stop by the posters and engage in more in-depth discussions with the presenters.

Sarra Habchi talks about challenges and opportunities in the quality engineering research in a video game industry #CSER2023 CSER Meetings - Canadian Software Eng. Research

Sarra Habchi an R&D scientist from Ubisoft is talking about "#QualityEngineering in the #videogame Industry: Challenges and Opportunities" at #CSERspr23 .

Presenting our work on Lifelong Learning for Software Analytics Models at #CSERspr23

I got the chance to participate in #FMSEVISION2030 the last few days and participate in discussions on the future of Foundation Models in the context of SE. Leaving with motivation, inspiration and ideas for the future of research on that subject. Thanks to the organizers!

I presented our work on the Cost and Benefits of Lifelong Learning for Software Analytics at #icse24 SEIP. We found that LL setups do as well, if not better, than traditional ML trainings in term of F1-score while reducting 2.3-12.5x the compute effort. shorturl.at/anpsH

At ICSE, the team not only showcased our research, they earned the Distinguished Paper Award of the ICSE SEIP track. Congratulations to the entire team for this well-deserved recognition ! 🏆 #ICSE #SEIP #SoftwareAnalytics #ubisoft arxiv.org/pdf/2305.09824 conf.researchr.org/info/icse-2024…

I presented our work in collaboration with La Forge on review activity prediction at #CSER2024. We found that ordering files based on review activity needs impacts review processes, motivating a prediction tool. This work was published at FSE24 arxiv.org/abs/2404.10703

File ordering has an impact on review processes. At FSE 2025, I talked about our paper's findings regarding ordering files in patches based on their likeliness of needing review. Read more about our paper in collaboration with Ubisoft Montréal La Forge: arxiv.org/abs/2404.10703