Learning to Learn with Contrastive Meta-Objective
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912692075233280 |
|---|---|
| author | Wu, Shiguang Wang, Yaqing Bian, Yatao Yao, Quanming |
| author_facet | Wu, Shiguang Wang, Yaqing Bian, Yatao Yao, Quanming |
| contents | Meta-learning enables learning systems to adapt quickly to new tasks, similar to humans. Different meta-learning approaches all work under/with the mini-batch episodic training framework. Such framework naturally gives the information about task identity, which can serve as additional supervision for meta-training to improve generalizability. We propose to exploit task identity as additional supervision in meta-training, inspired by the alignment and discrimination ability which is is intrinsic in human's fast learning. This is achieved by contrasting what meta-learners learn, i.e., model representations. The proposed ConML is evaluating and optimizing the contrastive meta-objective under a problem- and learner-agnostic meta-training framework. We demonstrate that ConML integrates seamlessly with existing meta-learners, as well as in-context learning models, and brings significant boost in performance with small implementation cost. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_05975 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning to Learn with Contrastive Meta-Objective Wu, Shiguang Wang, Yaqing Bian, Yatao Yao, Quanming Machine Learning Meta-learning enables learning systems to adapt quickly to new tasks, similar to humans. Different meta-learning approaches all work under/with the mini-batch episodic training framework. Such framework naturally gives the information about task identity, which can serve as additional supervision for meta-training to improve generalizability. We propose to exploit task identity as additional supervision in meta-training, inspired by the alignment and discrimination ability which is is intrinsic in human's fast learning. This is achieved by contrasting what meta-learners learn, i.e., model representations. The proposed ConML is evaluating and optimizing the contrastive meta-objective under a problem- and learner-agnostic meta-training framework. We demonstrate that ConML integrates seamlessly with existing meta-learners, as well as in-context learning models, and brings significant boost in performance with small implementation cost. |
| title | Learning to Learn with Contrastive Meta-Objective |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2410.05975 |