Vector Institute Research Featured at ICLR 2023 Conference
The Vector Institute for Artificial Intelligence announced that its faculty members and affiliates had 21 papers accepted at the 2023 International Conference on Learning Representations (ICLR), held from May 1 to May 5. The research highlights significant advancements in automated language processing, predictive AI, and reinforcement learning. Notable contributions include Jimmy Ba’s work on Automatic Prompt Engineering, which enables large language models to generate instructions and process human commands with greater precision, aiming for human-level performance in text generation. Additionally, Pascal Poupart’s paper on Inverse Constrained Reinforcement Learning (ICRL) introduces a new benchmark for modeling human-like behavior in autonomous driving and robot control by estimating constraints from expert demonstrations. These developments underscore the institute's role in advancing generative AI and safe deployment of reinforcement learning agents in physical systems. The conference served as a global platform for deep learning researchers to present novel algorithms and methodologies, reflecting ongoing progress in making AI systems more robust, interpretable, and aligned with human intentions.
Wire timeline
Vector Institute Research Featured at ICLR 2023 Conference
The Vector Institute for Artificial Intelligence announced that its faculty members and affiliates had 21 papers accepted at the 2023 International Conference on Learning Representations (ICLR), held from May 1 to May 5. The research highlights significant advancements in automated language processing, predictive AI, and reinforcement learning. Notable contributions include Jimmy Ba’s work on Automatic Prompt Engineering, which enables large language models to generate instructions and process human commands with greater precision, aiming for human-level performance in text generation. Additionally, Pascal Poupart’s paper on Inverse Constrained Reinforcement Learning (ICRL) introduces a new benchmark for modeling human-like behavior in autonomous driving and robot control by estimating constraints from expert demonstrations. These developments underscore the institute's role in advancing generative AI and safe deployment of reinforcement learning agents in physical systems. The conference served as a global platform for deep learning researchers to present novel algorithms and methodologies, reflecting ongoing progress in making AI systems more robust, interpretable, and aligned with human intentions.
Vector Institute for Artificial Intelligence