JasperoraProgram Guided AgentPeople can achieve goal by following natural language instructions. However, natural language is ambiguous and even people can feel…Aug 28, 2022Aug 28, 2022
JasperoraEfficient Off-Policy Meta-Reinforcement Learning via Probabilistic Context VariablesMost meta-RL methods need on-policy data during both meta-training and meta-testing. These methods lack efficiency when sampling, and…Aug 26, 2022Aug 26, 2022
JasperoraLanguage Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied AgentsLarge language models(LLMs) are trained using a colossal amount of data. Previous work shows that LLMs can internalize rich world…Aug 22, 20221Aug 22, 20221
JasperoraMultimodal Model-Agnostic Meta-Learning via Task-Aware Modulation ( MMAML )Model-Agnostic Meta-Learning (MAML) is a good way to make machine pick up a skill in a short time. This method relies on a single…Aug 19, 2022Aug 19, 2022
JasperoraModel-Agnostic Meta-Learning for Fast Adaptation of Deep Networks ( MAML )Human can learn a skill within a short time, but in artificial intelligence, agent can’t often do that. This can result from common sense…Aug 17, 20222Aug 17, 20222
JasperoraLeveraging Grammar And Reinforcement Learning For Neural Program SynthesisSince Neural Program Induction can’t explicitly output a program, Neural Program Synthesis is proposed to solve the problem. Neural Program…Aug 11, 20221Aug 11, 20221
JasperoraNeural Scene De-renderingAn encoding-decoding framework can be used to obtain interpretable representation of image. The framework utilize neural nets for both…Aug 11, 2022Aug 11, 2022
JasperoraLearning to Describe Scenes with ProgramsWhen people see a new image, we not only identify the size, color, position of every objects but also find the high-level relations like…Aug 9, 2022Aug 9, 2022
JasperoraDeep Q NetworkWhen talking about how to train agent to play games, some reinforcement learning applications use hand-crafted features to train model…Aug 5, 2022Aug 5, 2022
JasperoraProgrammatically Interpretable Reinforcement LearningIn Deep Reinforcement Learning(DRL), the policy is depicted as a neural network. If we can use program to represent policy, then it is more…Aug 4, 20221Aug 4, 20221