Improving the performance of weak supervision searches using transfer and meta-learning
Abstract Weak supervision searches have in principle the advantages of both being able to train on experimental data and being able to learn distinctive signal properties. However, the practical applicability of such searches is limited by the fact that successfully training a neural network via wea...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2024-02-01
|
| Series: | Journal of High Energy Physics |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/JHEP02(2024)138 |
| _version_ | 1797276059377336320 |
|---|---|
| author | Hugues Beauchesne Zong-En Chen Cheng-Wei Chiang |
| author_facet | Hugues Beauchesne Zong-En Chen Cheng-Wei Chiang |
| author_sort | Hugues Beauchesne |
| collection | DOAJ |
| description | Abstract Weak supervision searches have in principle the advantages of both being able to train on experimental data and being able to learn distinctive signal properties. However, the practical applicability of such searches is limited by the fact that successfully training a neural network via weak supervision can require a large amount of signal. In this work, we seek to create neural networks that can learn from less experimental signal by using transfer and meta-learning. The general idea is to first train a neural network on simulations, thereby learning concepts that can be reused or becoming a more efficient learner. The neural network would then be trained on experimental data and should require less signal because of its previous training. We find that transfer and meta-learning can substantially improve the performance of weak supervision searches. |
| first_indexed | 2024-03-07T15:22:53Z |
| format | Article |
| id | doaj.art-5ea7708076b74872ada4b40a16a2c31f |
| institution | Directory Open Access Journal |
| issn | 1029-8479 |
| language | English |
| last_indexed | 2024-03-07T15:22:53Z |
| publishDate | 2024-02-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Journal of High Energy Physics |
| spelling | doaj.art-5ea7708076b74872ada4b40a16a2c31f2024-03-05T17:29:26ZengSpringerOpenJournal of High Energy Physics1029-84792024-02-012024211910.1007/JHEP02(2024)138Improving the performance of weak supervision searches using transfer and meta-learningHugues Beauchesne0Zong-En Chen1Cheng-Wei Chiang2Physics Division, National Center for Theoretical SciencesDepartment of Physics and Center for Theoretical Physics, National Taiwan UniversityDepartment of Physics and Center for Theoretical Physics, National Taiwan UniversityAbstract Weak supervision searches have in principle the advantages of both being able to train on experimental data and being able to learn distinctive signal properties. However, the practical applicability of such searches is limited by the fact that successfully training a neural network via weak supervision can require a large amount of signal. In this work, we seek to create neural networks that can learn from less experimental signal by using transfer and meta-learning. The general idea is to first train a neural network on simulations, thereby learning concepts that can be reused or becoming a more efficient learner. The neural network would then be trained on experimental data and should require less signal because of its previous training. We find that transfer and meta-learning can substantially improve the performance of weak supervision searches.https://doi.org/10.1007/JHEP02(2024)138New Gauge InteractionsSpecific BSM Phenomenology |
| spellingShingle | Hugues Beauchesne Zong-En Chen Cheng-Wei Chiang Improving the performance of weak supervision searches using transfer and meta-learning Journal of High Energy Physics New Gauge Interactions Specific BSM Phenomenology |
| title | Improving the performance of weak supervision searches using transfer and meta-learning |
| title_full | Improving the performance of weak supervision searches using transfer and meta-learning |
| title_fullStr | Improving the performance of weak supervision searches using transfer and meta-learning |
| title_full_unstemmed | Improving the performance of weak supervision searches using transfer and meta-learning |
| title_short | Improving the performance of weak supervision searches using transfer and meta-learning |
| title_sort | improving the performance of weak supervision searches using transfer and meta learning |
| topic | New Gauge Interactions Specific BSM Phenomenology |
| url | https://doi.org/10.1007/JHEP02(2024)138 |
| work_keys_str_mv | AT huguesbeauchesne improvingtheperformanceofweaksupervisionsearchesusingtransferandmetalearning AT zongenchen improvingtheperformanceofweaksupervisionsearchesusingtransferandmetalearning AT chengweichiang improvingtheperformanceofweaksupervisionsearchesusingtransferandmetalearning |