Zero-shot Learning for Subdiscrimination in Pre-trained Models
In deep metric learning (DML) high-level input data are represented in a lower-level representation (embedding) space, such that samples from the same class are mapped close together, while samples from disparate classes are mapped further apart. In this lower-level representation, only a single inf...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Graz University of Technology
2025-01-01
|
| Series: | Journal of Universal Computer Science |
| Subjects: | |
| Online Access: | https://lib.jucs.org/article/120860/download/pdf/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832581893111939072 |
|---|---|
| author | Francisco Dominguez-Mateos Vincent O’Brien James Garland Ryan Furlong Daniel Palacios-Alonso |
| author_facet | Francisco Dominguez-Mateos Vincent O’Brien James Garland Ryan Furlong Daniel Palacios-Alonso |
| author_sort | Francisco Dominguez-Mateos |
| collection | DOAJ |
| description | In deep metric learning (DML) high-level input data are represented in a lower-level representation (embedding) space, such that samples from the same class are mapped close together, while samples from disparate classes are mapped further apart. In this lower-level representation, only a single inference sample from each known class is required to accurately discriminate between classes. To this end, embeddings trained for a specific task may contain additional feature information which can be used to go a level deeper into the discrimination task, i.e. allowing for feature sub-discrimination. This study takes an embedding trained to discriminate faces (identities) and uses the inherent feature information within the embedding to differentiate several attributes such as gender, age, and skin tone, without any additional training. This study is split into two cases; intra class discrimination where all the embeddings considered are from the same identity/in-dividual but with minor attributes such as beard/beardless, glasses/without glasses and emotions; and extra class discrimination where the embeddings represent different identities/people with more prominent attributes such as male/female, pale/dark tone, young/older. In the intra class sub-discriminant scenario, the inference process distinguishes common attributes and several artefacts of different identities, achieving 90.0% and 76.0% accuracy for beards and glasses, respectively. The system can also perform extra class sub-discrimination with a high accuracy rate, notably 99.3%, 99.3% and 94.1% for gender, skin tone, and age, respectively. To sum up, this work investigates the sub-discriminative capabilities of DML models by clustering discriminative features evident within the structure of DML embeddings. |
| format | Article |
| id | doaj-art-52c4c17d979f401ca1ff6f7a466e3242 |
| institution | Kabale University |
| issn | 0948-6968 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Graz University of Technology |
| record_format | Article |
| series | Journal of Universal Computer Science |
| spelling | doaj-art-52c4c17d979f401ca1ff6f7a466e32422025-01-30T08:31:23ZengGraz University of TechnologyJournal of Universal Computer Science0948-69682025-01-013119311010.3897/jucs.120860120860Zero-shot Learning for Subdiscrimination in Pre-trained ModelsFrancisco Dominguez-Mateos0Vincent O’Brien1James Garland2Ryan Furlong3Daniel Palacios-Alonso4University King Juan CarlosSouth East Technological UniversitySouth East Technological UniversitySouth East Technological UniversityUniversity King Juan CarlosIn deep metric learning (DML) high-level input data are represented in a lower-level representation (embedding) space, such that samples from the same class are mapped close together, while samples from disparate classes are mapped further apart. In this lower-level representation, only a single inference sample from each known class is required to accurately discriminate between classes. To this end, embeddings trained for a specific task may contain additional feature information which can be used to go a level deeper into the discrimination task, i.e. allowing for feature sub-discrimination. This study takes an embedding trained to discriminate faces (identities) and uses the inherent feature information within the embedding to differentiate several attributes such as gender, age, and skin tone, without any additional training. This study is split into two cases; intra class discrimination where all the embeddings considered are from the same identity/in-dividual but with minor attributes such as beard/beardless, glasses/without glasses and emotions; and extra class discrimination where the embeddings represent different identities/people with more prominent attributes such as male/female, pale/dark tone, young/older. In the intra class sub-discriminant scenario, the inference process distinguishes common attributes and several artefacts of different identities, achieving 90.0% and 76.0% accuracy for beards and glasses, respectively. The system can also perform extra class sub-discrimination with a high accuracy rate, notably 99.3%, 99.3% and 94.1% for gender, skin tone, and age, respectively. To sum up, this work investigates the sub-discriminative capabilities of DML models by clustering discriminative features evident within the structure of DML embeddings.https://lib.jucs.org/article/120860/download/pdf/Machine learningUnsupervised learningDeep metr |
| spellingShingle | Francisco Dominguez-Mateos Vincent O’Brien James Garland Ryan Furlong Daniel Palacios-Alonso Zero-shot Learning for Subdiscrimination in Pre-trained Models Journal of Universal Computer Science Machine learning Unsupervised learning Deep metr |
| title | Zero-shot Learning for Subdiscrimination in Pre-trained Models |
| title_full | Zero-shot Learning for Subdiscrimination in Pre-trained Models |
| title_fullStr | Zero-shot Learning for Subdiscrimination in Pre-trained Models |
| title_full_unstemmed | Zero-shot Learning for Subdiscrimination in Pre-trained Models |
| title_short | Zero-shot Learning for Subdiscrimination in Pre-trained Models |
| title_sort | zero shot learning for subdiscrimination in pre trained models |
| topic | Machine learning Unsupervised learning Deep metr |
| url | https://lib.jucs.org/article/120860/download/pdf/ |
| work_keys_str_mv | AT franciscodominguezmateos zeroshotlearningforsubdiscriminationinpretrainedmodels AT vincentobrien zeroshotlearningforsubdiscriminationinpretrainedmodels AT jamesgarland zeroshotlearningforsubdiscriminationinpretrainedmodels AT ryanfurlong zeroshotlearningforsubdiscriminationinpretrainedmodels AT danielpalaciosalonso zeroshotlearningforsubdiscriminationinpretrainedmodels |