Vector Institute Researchers Present Over 50 Papers at NeurIPS 2021
Researchers from the Vector Institute for Artificial Intelligence are presenting more than 50 papers at the virtual Conference on Neural Information Processing Systems (NeurIPS) 2021, held online from December 6 to 14. The contributions span diverse AI fields, including deep learning, reinforcement learning, computer vision, and responsible AI. Notable works include a collaboration between Graham Taylor’s team and POSTECH researchers on using deep reinforcement learning for combinatorial LEGO construction, which has potential applications in architectural design. Another significant study by Richard Zemel and Alireza Makhzani addresses neural network vulnerabilities to model inversion attacks, aiming to enhance machine learning privacy, particularly in healthcare. Additionally, Chris Maddison and colleagues introduced a novel data compression paradigm that outperforms JPEG significantly, potentially lowering compute costs for smaller institutions. These advancements highlight Vector’s ongoing commitment to breaking new ground in AI research with practical implications for daily life and industry.
Wire timeline
Vector Institute Researchers Present Over 50 Papers at NeurIPS 2021
Researchers from the Vector Institute for Artificial Intelligence are presenting more than 50 papers at the virtual Conference on Neural Information Processing Systems (NeurIPS) 2021, held online from December 6 to 14. The contributions span diverse AI fields, including deep learning, reinforcement learning, computer vision, and responsible AI. Notable works include a collaboration between Graham Taylor’s team and POSTECH researchers on using deep reinforcement learning for combinatorial LEGO construction, which has potential applications in architectural design. Another significant study by Richard Zemel and Alireza Makhzani addresses neural network vulnerabilities to model inversion attacks, aiming to enhance machine learning privacy, particularly in healthcare. Additionally, Chris Maddison and colleagues introduced a novel data compression paradigm that outperforms JPEG significantly, potentially lowering compute costs for smaller institutions. These advancements highlight Vector’s ongoing commitment to breaking new ground in AI research with practical implications for daily life and industry.
Vector Institute for Artificial Intelligence