Knowledge Extraction with Deep Belief Networks
Thesis title in Czech: | Extrakce znalostí pomocí DBN-sítí |
---|---|
Thesis title in English: | Knowledge Extraction with Deep Belief Networks |
Key words: | DBN-sítě|RBM-sítě|extrakce znalostí|reprezentace znalostí|prořezávání|optimalizace architektury |
English key words: | deep belief networks|restricted Boltzmann machines|knowledge extraction|knowledge representation|pruning|architecture optimization |
Academic year of topic announcement: | 2022/2023 |
Thesis type: | Bachelor's thesis |
Thesis language: | angličtina |
Department: | Department of Theoretical Computer Science and Mathematical Logic (32-KTIML) |
Supervisor: | doc. RNDr. Iveta Mrázová, CSc. |
Author: | Bc. Jan Bronec - assigned and confirmed by the Study Dept. |
Date of registration: | 20.01.2023 |
Date of assignment: | 24.01.2023 |
Confirmed by Study dept. on: | 15.02.2023 |
Date and time of defence: | 29.06.2023 09:00 |
Date of electronic submission: | 10.05.2023 |
Date of submission of printed version: | 10.05.2023 |
Date of proceeded defence: | 29.06.2023 |
Opponents: | RNDr. Věra Flídrová |
Guidelines |
The student shall review the following topics in his thesis:
- recapitulation of the paradigms applicable to training of deep belief networks, in particular, RBM-networks, contrastive divergence, DBN-networks, and fine-tuning, - overview and mutual comparison of various approaches applicable to knowledge extraction with neural networks, e.g., rule extraction/insertion, pruning, and visualization using UMAP, The student will focus on some of these topics in more detail. Further, he will propose a suitable strategy for reliable classification of the presented objects, e.g., real-world image data, and will implement the models. Evaluating the obtained results and the gained experience shall form an essential part of the thesis. |
References |
1. Some of the textbooks available for the chosen area of research, e.g.:
- C. C. Aggarwal: Neural Networks and Deep Learning: A Textbook, Springer (2018). - S. Marsland: Machine Learning: An Algorithmic Perspective, 2nd Edition, Chapman & Hall/CRC (2015). 2. Journal papers and other publications: - S.-K. Chao, Z. Wang, Y. Xing, and G. Cheng: Directional Pruning of Deep Neural Networks, in: Advances in Neural Information Processing Systems, vol. 33, Curran Associates, Inc. (2020), pp. 13986--13998. - X. Dong, S. Chen, and S. Pan: Learning to Prune Deep Neural Networks via Layer-wise Optimal Brain Surgeon, in: Advances in Neural Information Processing Systems, vol. 30, Curran Associates, Inc. (2017), 11 p. - T. Hoefler, D. Alistarh, T. Ben-Nun, N. Dryden, and A. Peste: Sparsity in Deep Learning: Pruning and growth for efficient inference and training in neural networks, in: Journal of Machine Learning Research, vol. 23 (2021), pp. 1-124. - L. McInnes, J. Healy, and J. Melville: UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction, in: arXiv:1802.03426v3, (2020). - M. Pietron, M. Wielgosz: Retrain or Not Retrain? - Efficient Pruning Methods of Deep CNN Networks, in: LNCS 12139 (2020), pp. 452-463. - S. N. Tran, A. S. d´Avila Garcez: Deep Logic Networks: Inserting and Extracting Knowledge From Deep Belief Network, in: IEEE Transactions on Neural Networks and Learning Systems, vol. 29, no. 2 (Feb. 2018), pp. 246-258. - H. Wang, C. Qin, Y. Bai, Y. Zhang, and Y. Fu: Recent Advances on Neural Network Pruning at Initialization, in: Proc. of IJCAI-22, IJCAI Organization (2022), pp.5638-5645. 3. Relevant articles from leading academic journals: Neurocomputing, Neural Networks, IEEE Transactions on Neural Networks and Learning Systems, etc. |