Uplifting Lower-Income Data: Strategies for Socioeconomic Perspective Shifts in Large Multi-modal Models
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_ | 1866914971135246336 |
|---|---|
| author | Nwatu, Joan Ignat, Oana Mihalcea, Rada |
| author_facet | Nwatu, Joan Ignat, Oana Mihalcea, Rada |
| contents | Recent work has demonstrated that the unequal representation of cultures and socioeconomic groups in training data leads to biased Large Multi-modal (LMM) models. To improve LMM model performance on underrepresented data, we propose and evaluate several prompting strategies using non-English, geographic, and socioeconomic attributes. We show that these geographic and socioeconomic integrated prompts favor retrieving topic appearances commonly found in data from low-income households across different countries leading to improved LMM model performance on lower-income data. Our analyses identify and highlight contexts where these strategies yield the most improvements. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_02623 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Uplifting Lower-Income Data: Strategies for Socioeconomic Perspective Shifts in Large Multi-modal Models Nwatu, Joan Ignat, Oana Mihalcea, Rada Computers and Society Artificial Intelligence Computation and Language Computer Vision and Pattern Recognition K.4; I.2.7; I.2.8 Recent work has demonstrated that the unequal representation of cultures and socioeconomic groups in training data leads to biased Large Multi-modal (LMM) models. To improve LMM model performance on underrepresented data, we propose and evaluate several prompting strategies using non-English, geographic, and socioeconomic attributes. We show that these geographic and socioeconomic integrated prompts favor retrieving topic appearances commonly found in data from low-income households across different countries leading to improved LMM model performance on lower-income data. Our analyses identify and highlight contexts where these strategies yield the most improvements. |
| title | Uplifting Lower-Income Data: Strategies for Socioeconomic Perspective Shifts in Large Multi-modal Models |
| topic | Computers and Society Artificial Intelligence Computation and Language Computer Vision and Pattern Recognition K.4; I.2.7; I.2.8 |
| url | https://arxiv.org/abs/2407.02623 |