Welcome to ACM SIGEnergy Seminar website...!!!
ACM SIGEnergy Seminar is a monthly virtual seminar series for graduate students, postdocs and junior faculty to present their ongoing work to peers and mentors around the world. The seminar series provides an opportunity for knowledge exchange in our community in between our conferences e-Energy and BuildSys.
Join our special session on Dec. 13th at 11:00 am ET (Zoom)
Abstract:
Reinforcement learning (RL) offers a promising approach to optimize building energy systems, but existing simulation environments present scalability and efficiency challenges. This keynote presents an open-source, physics-based building simulator designed for rapid RL agent training. We detail the simulator's lightweight architecture, featuring a finite-difference solver and simplified HVAC model, and highlight a novel matrix-based implementation that achieves a 10x speedup through TensorFlow computation. We discuss the open-source dataset of building telemetry accompanying the simulator and outline future enhancements, including advanced heat transfer models and improved agent design. Finally, we invite collaboration with researchers and developers to expand the simulator's capabilities and accelerate innovation in sustainable building technologies.
12/13/2024: ACM SIGEnergy seminar special session featuring John Sipple from Google Research.
11/27/2024: We concluded the final seminar of the 2024 cycle.
08/15/2024: We achieved record attendance in our July seminar. Stay tuned for the upcoming seminars!
04/24/2024: Our first seminar in 2024 was a success. Big thanks to those who attended and presented!
03/24/2023: Kingsley Nweye and Walid Hanafy will be presenting their work in the third round of talks on 03/29/2023 at 08:00AM Pacific Time. .
01/17/2023: Seminar website is launched and first set of talks are scheduled for 01/25/2023 at 08:00AM Pacific Time.
01/17/2023: Ankita Garg and Ahmed S. Alahmed will be presenting their work in the first seminar.