Zobrazeno 1 - 10
of 152
pro vyhledávání: '"Pilly, Praveen"'
Autor:
Valiente, Rodolfo, Pilly, Praveen K.
Metacognition--the awareness and regulation of one's cognitive processes--is central to human adaptability in unknown situations. In contrast, current autonomous agents often struggle in novel environments due to their limited capacity for adaptation
Externí odkaz:
http://arxiv.org/abs/2411.13537
Autor:
Soltoggio, Andrea, Ben-Iwhiwhu, Eseoghene, Peridis, Christos, Ladosz, Pawel, Dick, Jeffery, Pilly, Praveen K., Kolouri, Soheil
This paper introduces a set of formally defined and transparent problems for reinforcement learning algorithms with the following characteristics: (1) variable degrees of observability (non-Markov observations), (2) distal and sparse rewards, (3) var
Externí odkaz:
http://arxiv.org/abs/2302.10887
Autor:
Baker, Megan M., New, Alexander, Aguilar-Simon, Mario, Al-Halah, Ziad, Arnold, Sébastien M. R., Ben-Iwhiwhu, Ese, Brna, Andrew P., Brooks, Ethan, Brown, Ryan C., Daniels, Zachary, Daram, Anurag, Delattre, Fabien, Dellana, Ryan, Eaton, Eric, Fu, Haotian, Grauman, Kristen, Hostetler, Jesse, Iqbal, Shariq, Kent, Cassandra, Ketz, Nicholas, Kolouri, Soheil, Konidaris, George, Kudithipudi, Dhireesha, Learned-Miller, Erik, Lee, Seungwon, Littman, Michael L., Madireddy, Sandeep, Mendez, Jorge A., Nguyen, Eric Q., Piatko, Christine D., Pilly, Praveen K., Raghavan, Aswin, Rahman, Abrar, Ramakrishnan, Santhosh Kumar, Ratzlaff, Neale, Soltoggio, Andrea, Stone, Peter, Sur, Indranil, Tang, Zhipeng, Tiwari, Saket, Vedder, Kyle, Wang, Felix, Xu, Zifan, Yanguas-Gil, Angel, Yedidsion, Harel, Yu, Shangqun, Vallabha, Gautam K.
Despite the advancement of machine learning techniques in recent years, state-of-the-art systems lack robustness to "real world" events, where the input distributions and tasks encountered by the deployed systems will not be limited to the original t
Externí odkaz:
http://arxiv.org/abs/2301.07799
Autor:
Ben-Iwhiwhu, Eseoghene, Nath, Saptarshi, Pilly, Praveen K., Kolouri, Soheil, Soltoggio, Andrea
Publikováno v:
Transactions on Machine Learning Research (2023)
Lifelong learning aims to create AI systems that continuously and incrementally learn during a lifetime, similar to biological learning. Attempts so far have met problems, including catastrophic forgetting, interference among tasks, and the inability
Externí odkaz:
http://arxiv.org/abs/2212.11110
Autor:
Ketz, Nicholas A., Pilly, Praveen K.
The robustness of any machine learning solution is fundamentally bound by the data it was trained on. One way to generalize beyond the original training is through human-informed augmentation of the original dataset; however, it is impossible to spec
Externí odkaz:
http://arxiv.org/abs/2209.03207
Reinforcement learning agents perform well when presented with inputs within the distribution of those encountered during training. However, they are unable to respond effectively when faced with novel, out-of-distribution events, until they have und
Externí odkaz:
http://arxiv.org/abs/2112.09670
Autor:
Ben-Iwhiwhu, Eseoghene, Dick, Jeffery, Ketz, Nicholas A., Pilly, Praveen K., Soltoggio, Andrea
Publikováno v:
Neural Networks 2022
Meta-reinforcement learning (meta-RL) algorithms enable agents to adapt quickly to tasks from few samples in dynamic environments. Such a feat is achieved through dynamic representations in an agent's policy network (obtained via reasoning about task
Externí odkaz:
http://arxiv.org/abs/2111.00134
Autor:
Li, Haoran, Krishnan, Aditya, Wu, Jingfeng, Kolouri, Soheil, Pilly, Praveen K., Braverman, Vladimir
Preventing catastrophic forgetting while continually learning new tasks is an essential problem in lifelong learning. Structural regularization (SR) refers to a family of algorithms that mitigate catastrophic forgetting by penalizing the network for
Externí odkaz:
http://arxiv.org/abs/2104.08604
Autor:
Ben-Iwhiwhu, Eseoghene, Ladosz, Pawel, Dick, Jeffery, Chen, Wen-Hua, Pilly, Praveen, Soltoggio, Andrea
Rapid online adaptation to changing tasks is an important problem in machine learning and, recently, a focus of meta-reinforcement learning. However, reinforcement learning (RL) algorithms struggle in POMDP environments because the state of the syste
Externí odkaz:
http://arxiv.org/abs/2004.12846
Autor:
Ladosz, Pawel, Ben-Iwhiwhu, Eseoghene, Dick, Jeffery, Hu, Yang, Ketz, Nicholas, Kolouri, Soheil, Krichmar, Jeffrey L., Pilly, Praveen, Soltoggio, Andrea
This paper presents a new neural architecture that combines a modulated Hebbian network (MOHN) with DQN, which we call modulated Hebbian plus Q network architecture (MOHQA). The hypothesis is that such a combination allows MOHQA to solve difficult pa
Externí odkaz:
http://arxiv.org/abs/1909.09902