Learning Embeddings for Sequential Tasks Using Population of Agents
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2023
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916239395258368 |
|---|---|
| author | Mahajan, Mridul Tzannetos, Georgios Radanovic, Goran Singla, Adish |
| author_facet | Mahajan, Mridul Tzannetos, Georgios Radanovic, Goran Singla, Adish |
| contents | We present an information-theoretic framework to learn fixed-dimensional embeddings for tasks in reinforcement learning. We leverage the idea that two tasks are similar if observing an agent's performance on one task reduces our uncertainty about its performance on the other. This intuition is captured by our information-theoretic criterion which uses a diverse agent population as an approximation for the space of agents to measure similarity between tasks in sequential decision-making settings. In addition to qualitative assessment, we empirically demonstrate the effectiveness of our techniques based on task embeddings by quantitative comparisons against strong baselines on two application scenarios: predicting an agent's performance on a new task by observing its performance on a small quiz of tasks, and selecting tasks with desired characteristics from a given set of options. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2306_03311 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Learning Embeddings for Sequential Tasks Using Population of Agents Mahajan, Mridul Tzannetos, Georgios Radanovic, Goran Singla, Adish Machine Learning Artificial Intelligence We present an information-theoretic framework to learn fixed-dimensional embeddings for tasks in reinforcement learning. We leverage the idea that two tasks are similar if observing an agent's performance on one task reduces our uncertainty about its performance on the other. This intuition is captured by our information-theoretic criterion which uses a diverse agent population as an approximation for the space of agents to measure similarity between tasks in sequential decision-making settings. In addition to qualitative assessment, we empirically demonstrate the effectiveness of our techniques based on task embeddings by quantitative comparisons against strong baselines on two application scenarios: predicting an agent's performance on a new task by observing its performance on a small quiz of tasks, and selecting tasks with desired characteristics from a given set of options. |
| title | Learning Embeddings for Sequential Tasks Using Population of Agents |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2306.03311 |