Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2410.15875 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910658171240448 |
|---|---|
| author | Graffeuille, Olivier Koh, Yun Sing Wicker, Joerg Lehmann, Moritz |
| author_facet | Graffeuille, Olivier Koh, Yun Sing Wicker, Joerg Lehmann, Moritz |
| contents | Knowledge transfer in multi-task learning is typically viewed as a dichotomy; positive transfer, which improves the performance of all tasks, or negative transfer, which hinders the performance of all tasks. In this paper, we investigate the understudied problem of asymmetric task relationships, where knowledge transfer aids the learning of certain tasks while hindering the learning of others. We propose an optimisation strategy that includes additional cloned tasks named self-auxiliaries into the learning process to flexibly transfer knowledge between tasks asymmetrically. Our method can exploit asymmetric task relationships, benefiting from the positive transfer component while avoiding the negative transfer component. We demonstrate that asymmetric knowledge transfer provides substantial improvements in performance compared to existing multi-task optimisation strategies on benchmark computer vision problems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_15875 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Enabling Asymmetric Knowledge Transfer in Multi-Task Learning with Self-Auxiliaries Graffeuille, Olivier Koh, Yun Sing Wicker, Joerg Lehmann, Moritz Machine Learning Knowledge transfer in multi-task learning is typically viewed as a dichotomy; positive transfer, which improves the performance of all tasks, or negative transfer, which hinders the performance of all tasks. In this paper, we investigate the understudied problem of asymmetric task relationships, where knowledge transfer aids the learning of certain tasks while hindering the learning of others. We propose an optimisation strategy that includes additional cloned tasks named self-auxiliaries into the learning process to flexibly transfer knowledge between tasks asymmetrically. Our method can exploit asymmetric task relationships, benefiting from the positive transfer component while avoiding the negative transfer component. We demonstrate that asymmetric knowledge transfer provides substantial improvements in performance compared to existing multi-task optimisation strategies on benchmark computer vision problems. |
| title | Enabling Asymmetric Knowledge Transfer in Multi-Task Learning with Self-Auxiliaries |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2410.15875 |