TY - GEN
T1 - Job-Profile Matching with CTN and MADRL with GEABB: A Recommender System
AU - Gaddam, Jyotheesh
AU - Barca, Jan Carlo
AU - Nguven, Thanh Thi
AU - Angelova, Maia
PY - 2024/3/14
Y1 - 2024/3/14
N2 - In this paper, we propose a novel approach to address the complex and challenging task of job-profile matching by enhancing the performance of the co-teaching network model, which is a state-of-the-art deep learning-based approach. Our approach involves implementing a self-adaptable algorithm that integrates an optimisation layer and adaptive building blocks. The proposed algorithm is trained and tested on a Kaggle data-set, and our numerical experiment demonstrates that the integration layer improved the accuracy of the co-teaching network by 5%, significantly enhancing the performance of the model in job-profile matching with real-world data. Our approach also incorporates self-adaptive building blocks that can adapt to changing environments and improve prediction accuracy, and the integration of multi-agent deep reinforcement learning, particle swarm optimisation, and grammatical evolution enhances the optimisation process and improves the accuracy of the model. Overall, our study presents a novel integration of adaptive building blocks and multi-agent deep reinforcement learning with the co-teaching network for job-profile matching and shows that our approach can significantly improve the performance of existing state-of-the-art models in this area.
AB - In this paper, we propose a novel approach to address the complex and challenging task of job-profile matching by enhancing the performance of the co-teaching network model, which is a state-of-the-art deep learning-based approach. Our approach involves implementing a self-adaptable algorithm that integrates an optimisation layer and adaptive building blocks. The proposed algorithm is trained and tested on a Kaggle data-set, and our numerical experiment demonstrates that the integration layer improved the accuracy of the co-teaching network by 5%, significantly enhancing the performance of the model in job-profile matching with real-world data. Our approach also incorporates self-adaptive building blocks that can adapt to changing environments and improve prediction accuracy, and the integration of multi-agent deep reinforcement learning, particle swarm optimisation, and grammatical evolution enhances the optimisation process and improves the accuracy of the model. Overall, our study presents a novel integration of adaptive building blocks and multi-agent deep reinforcement learning with the co-teaching network for job-profile matching and shows that our approach can significantly improve the performance of existing state-of-the-art models in this area.
UR - https://ieeexplore.ieee.org/document/10569707
UR - http://www.scopus.com/inward/record.url?scp=85198341377&partnerID=8YFLogxK
U2 - 10.1109/iccae59995.2024.10569707
DO - 10.1109/iccae59995.2024.10569707
M3 - Conference publication
T3 - 2024 16th International Conference on Computer and Automation Engineering (ICCAE)
SP - 203
EP - 210
BT - 2024 16th International Conference on Computer and Automation Engineering (ICCAE)
PB - IEEE
ER -