Performance Evaluation of Deep Learning Models: A Review Based on The F2 Score

Mountassir Bay Tamsamani

Article Details

Performance Evaluation of Deep Learning Models: A Review Based on The F2 Score

Authors:

Mountassir Bay Tamsamani

Publisher:

School of Information Sciences

Journal:

Journal of Information Sciences Vol.24 Issue 1, pp.120-144

Document Type:

Article

Publication Date:

8/2025

Document Language:

English

Document ID (Org No.):

2001845053

pISSN:

2820-6894

Abstract:

 Emotion detection is a key area in Natural Language Processing (NLP), with applications ranging from recommendation systems to conversational agents based on the capabilities of large language models. This paper evaluates the effectiveness of deep learning models for emotion detection in a textual corpus. From this dataset, we compare the performance of models based on Deep Learning architectures, evaluated using the F2 score metric. The results show the crucial role of the F2 score in the evaluation of classification models, especially in cases where false negatives can have a more decisive impact than false positives. By opting for the F2 score rather than the F1 score, our analysis, which is fully aware of the limitations of Artificial Intelligence in the broad sense, proposes an alternative approach to the evaluation of emotion detection systems, thereby widening the scope of research aimed at improving their consistency and fairness, especially in critical areas such as psychological assistance using conversational agents that generate natural text. Keywords: Emotion detection; Natural language Processing (NLP); Conversational agents; Deep Learning; F2-score; Artificial Intelligence.

Article Details

Performance Evaluation of Deep Learning Models: A Review Based on The F2 Score

Authors: Mountassir Bay Tamsamani

Publisher: School of Information Sciences

Journal: Journal of Information Sciences Vol.24 Issue 1, pp.120-144

Document Type: Article

Publication Date: 8/2025

Document Language: English

Document ID (Org No.): 2001845053

pISSN: 2820-6894

Issues

Authors:

Mountassir Bay Tamsamani

Publisher:

School of Information Sciences

Journal:

Journal of Information Sciences Vol.24 Issue 1, pp.120-144

Document Type:

Article

Publication Date:

8/2025

Document Language:

English

Document ID (Org No.):

2001845053

pISSN:

2820-6894