Fusion of transfer learning with nature-inspired dandelion algorithm for autism spectrum disorder detection and classification using facial features DOI Creative Commons

G. Elangovan,

N. Jagadish Kumar,

J. Shobana

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Dec. 28, 2024

Autism spectrum disorder (ASD) is a neurologic considered to cause discrepancies in physical activities, social skills, and cognition. There no specific medicine for treating this disorder; early intervention critical improving brain function. Additionally, the lack of clinical test detecting ASD makes diagnosis challenging. To regulate identification, physicians entertain children's activities growing histories. The human face employed as biological signature it has potential reflections brain. It utilized simpler more helpful tool detection. Artificial intelligence (AI) algorithms medicinal rehabilitation can help specialists identify various illnesses successfully. However, owing its particular heterogeneous symptoms complex nature, remains be challenging investigators. This work presents Fusion Transfer Learning (TL) with Dandelion Algorithm Accurate Spectrum Disorder Detection Classification (FTLDA-AASDDC) method. FTLDA-AASDDC technique detects classifies autism non-autism samples using facial images. accomplish this, utilizes bilateral filter (BF) approach noise elimination. Next, employs fusion-based TL process comprising three models, namely MobileNetV2, DenseNet201, ResNet50. Moreover, attention-based bi-directional long short-term memory (A-BiLSTM) method used classify recognize ASD. Finally, (DA) optimize parameter tuning process, efficacy A-BiLSTM technique. A wide range simulation analyses performed highlight classification performance experimental validation portrayed superior accuracy value 97.50% over existing techniques.

Language: Английский

Fusion of transfer learning with nature-inspired dandelion algorithm for autism spectrum disorder detection and classification using facial features DOI Creative Commons

G. Elangovan,

N. Jagadish Kumar,

J. Shobana

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Dec. 28, 2024

Autism spectrum disorder (ASD) is a neurologic considered to cause discrepancies in physical activities, social skills, and cognition. There no specific medicine for treating this disorder; early intervention critical improving brain function. Additionally, the lack of clinical test detecting ASD makes diagnosis challenging. To regulate identification, physicians entertain children's activities growing histories. The human face employed as biological signature it has potential reflections brain. It utilized simpler more helpful tool detection. Artificial intelligence (AI) algorithms medicinal rehabilitation can help specialists identify various illnesses successfully. However, owing its particular heterogeneous symptoms complex nature, remains be challenging investigators. This work presents Fusion Transfer Learning (TL) with Dandelion Algorithm Accurate Spectrum Disorder Detection Classification (FTLDA-AASDDC) method. FTLDA-AASDDC technique detects classifies autism non-autism samples using facial images. accomplish this, utilizes bilateral filter (BF) approach noise elimination. Next, employs fusion-based TL process comprising three models, namely MobileNetV2, DenseNet201, ResNet50. Moreover, attention-based bi-directional long short-term memory (A-BiLSTM) method used classify recognize ASD. Finally, (DA) optimize parameter tuning process, efficacy A-BiLSTM technique. A wide range simulation analyses performed highlight classification performance experimental validation portrayed superior accuracy value 97.50% over existing techniques.

Language: Английский

Citations

0