Výsledky vyhledávání - "Nakano, Akihiro"

Report

Object-Centric Temporal Consistency via Conditional Autoregressive Inductive Biases

Autor: Meo, Cristian, Nakano, Akihiro, Lică, Mircea, Didolkar, Aniket, Suzuki, Masahiro, Goyal, Anirudh, Zhang, Mengmi, Dauwels, Justin, Matsuo, Yutaka, Bengio, Yoshua

Unsupervised object-centric learning from videos is a promising approach towards learning compositional representations that can be applied to various downstream tasks, such as prediction and reasoning. Recently, it was shown that pretrained Vision T

Externí odkaz: http://arxiv.org/abs/2410.15728

Zobrazit plný text záznamu

Report

Masked Generative Priors Improve World Models Sequence Modelling Capabilities

Autor: Meo, Cristian, Lica, Mircea, Ikram, Zarif, Nakano, Akihiro, Shah, Vedant, Didolkar, Aniket Rajiv, Liu, Dianbo, Goyal, Anirudh, Dauwels, Justin

Deep Reinforcement Learning (RL) has become the leading approach for creating artificial agents in complex environments. Model-based approaches, which are RL methods with world models that predict environment dynamics, are among the most promising di

Externí odkaz: http://arxiv.org/abs/2410.07836

Zobrazit plný text záznamu

Report

Cross-Task Consistency Learning Framework for Multi-Task Learning

Autor: Nakano, Akihiro, Chen, Shi, Demachi, Kazuyuki

Multi-task learning (MTL) is an active field in deep learning in which we train a model to jointly learn multiple tasks by exploiting relationships between the tasks. It has been shown that MTL helps the model share the learned features between tasks

Externí odkaz: http://arxiv.org/abs/2111.14122

Zobrazit plný text záznamu