Emotion recognition using EEG Signals based on optimized deep Classifiers

dc.contributor.guideSHIRISH S. KULKARNI
dc.coverage.spatialDeep classifiers
dc.creator.researcherKHUBANI JITENDRA ANAND
dc.date.accessioned2024-08-30T03:43:10Z
dc.date.available2024-08-30T03:43:10Z
dc.date.awarded2024
dc.date.completed2024
dc.date.registered2020
dc.description.abstractEmotion identification is the current research concept as it obtains a significant role in interpersonal relationships and health care services. Even though a lot of work is concentrated on emotion recognition they still suffer from various issues such as noisy and inconsistent EEG signals, high testing and training time, and low training efficiency. newlineThis research presents emotion recognition with EEG data to progress the precise identification with input datasets such as DEAP and SEED. 40 electrodes are chosen from the DEAP dataset to facilitate effective emotion recognition. Similarly, around 40 electrodes are selected from the SEED dataset. newlineBrainstorm Optimization and Group Search Optimizer Salgorithms are introduced to optimize artificial neural networks (ANNs), enhancing electrode selection and parameter tuning for improved accuracy. The GSO-NN, after being fine-tuned, achieves 94.58% accuracy on the DEAP dataset and 91.92% on the SEED dataset. The BSO-NN, after being fine-tuned, achieves 93.95% accuracy on the DEAP dataset and 90.93% on the SEED dataset. The deep convolutional neural network (DCNN) architecture incorporates a novel brain algorithm for electrode selection and feature extraction. The inventive DCNN, after being fine-tuned which achieves 97.12% accuracy on the DEAP dataset and 94.71% on the SEED dataset. Additionally, an Adaptive Search Optimizer (ASO) is introduced to fine-tune the hyperparameters of deep Bi-LSTM models, further enhancing emotion recognition performance. The BiLSTM-ASO, after being fine-tuned, achieves 97.51% accuracy on the DEAP dataset and 97.47% on the SEED dataset. Furthermore, a Communicative-Cooperative and Attentive behavioral (C2A) optimization is employed to effectively select informative electrodes and fine-tune the attention-deep model (ADM), thereby enhancing emotion recognition performance. with minimal complexity. The C2A-ADM, after being fine-tuned using the C2A optimization, achieves 98.09% accuracy on the DEAP dataset and 98.31% on the SEED dataset. newline
dc.description.note
dc.format.accompanyingmaterialNone
dc.format.dimensions11.5MB
dc.format.extent189
dc.identifier.urihttp://hdl.handle.net/10603/586388
dc.languageEnglish
dc.publisher.institutionSchool of Engineering
dc.publisher.placeNavi Mumbai
dc.publisher.universityPadmashree Dr. D.Y. Patil Vidyapeeth, Navi Mumbai
dc.relationhttps://doi.org/10.1007/s13278-023-01035-6, https://doi.org/10.1016/j.bspc.2024.106217
dc.rightsuniversity
dc.source.universityUniversity
dc.subject.keywordEngineering
dc.subject.keywordEngineering and Technology
dc.subject.keywordInstruments and Instrumentation
dc.titleEmotion recognition using EEG Signals based on optimized deep Classifiers
dc.title.alternative
dc.type.degreePh.D.

Files

Original bundle

Now showing 1 - 5 of 14
Loading...
Thumbnail Image
Name:
01_title.pdf
Size:
108.81 KB
Format:
Adobe Portable Document Format
Description:
Attached File
Loading...
Thumbnail Image
Name:
02_prelim pages.pdf
Size:
529.31 KB
Format:
Adobe Portable Document Format
Loading...
Thumbnail Image
Name:
03_content.pdf
Size:
243.47 KB
Format:
Adobe Portable Document Format
Loading...
Thumbnail Image
Name:
04_abstract.pdf
Size:
88.7 KB
Format:
Adobe Portable Document Format
Loading...
Thumbnail Image
Name:
05_chapter 1.pdf
Size:
359.96 KB
Format:
Adobe Portable Document Format

License bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
license.txt
Size:
1.79 KB
Format:
Plain Text
Description: