Adversarial Patterns in Deep Belief Networks
Název práce v češtině: | Adversarial Patterns in Deep Belief Networks |
---|---|
Název v anglickém jazyce: | Adversarial Patterns in Deep Belief Networks |
Klíčová slova: | klasifikace|klastrování|umělé neuronové sítě|DBN-sítě|reprezentace znalostí |
Klíčová slova anglicky: | classification|clustering|artificial neural networks|DBN-networks|knowledge representation |
Akademický rok vypsání: | 2024/2025 |
Typ práce: | diplomová práce |
Jazyk práce: | |
Ústav: | Katedra teoretické informatiky a matematické logiky (32-KTIML) |
Vedoucí / školitel: | doc. RNDr. Iveta Mrázová, CSc. |
Řešitel: |
Zásady pro vypracování |
The student shall review the following topics in his/her diploma thesis:
- overview and comparison of various paradigms applicable to training of deep belief networks (e.g., gradient descent, contrastive divergence, restricted Boltzmann machines, deep belief networks, etc.), - recapitulation and mutual comparison of known approaches suitable for the detection of adversarial patterns, i.e., misclassified patterns that are, however, very similar to correctly classified patterns, (t-SNE embedding, rule extraction, and inference, among others), - interpretation and visualization of the found adversarial patterns and their characteristics. The student will focus on some of these topics in more detail. Further, he/she will propose a suitable strategy for the detection, generation, and utilization of adversarial patterns based on real-world data, e.g., images, and will implement the models. The evaluation of the obtained results and gained experience shall form an important part of the thesis. |
Seznam odborné literatury |
1. Některé z dostupných základních učebnic, resp. přehledových článků vhodných pro zvolené téma, např.:
- N. Buduma: Fundamentals of Deep Learning: Designing Next-Generation Machine Intelligence Algorithms, O´Reilly, (2017). - I. Goodfellow, Y. Bengio, and A. Courville: Deep Learning, The MIT Press, (2016). - S. Haykin: Neural Networks and Learning Machines, 3rd edition, Pearson, (2009). 2. Články: - D. H. Ackley, G. E. Hinton, and T. J. Sejnowski: A learning algorithm for Boltzmann machines, in: Cognitive science, Vol. 9(1), (1985), pp.147-169. - I. Goodfellow, J. Shlens, and C. Szegedy: Explaining and Harnessing Adversarial Examples, in: Proc. of ICLR´2015, (2015), CoRR, abs/1412.6572, 12 p. - G. E. Hinton: Training products of experts by minimizing contrastive divergence, in: Neural Computation, Vol. 14(8), (2002), pp. 1771-1800. - G. E. Hinton, S. Osindero, Y.-W. Teh: A Fast Learning Algorithm for Deep Belief Networks, in: Neural Computation, Vol. 18, (2006), pp. 1527-1554. - A. Fischer and C. Igel: An introduction to restricted Boltzmann machines, in: Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications, Springer, (2012), pp. 14-36. - N. Le Roux and Y. Bengio: Representational power of restricted Boltzmann machines and deep belief networks, in: Neural Computation, Vol. 20(6), (2008), pp. 1631-1649. - L. van der Maaten and G. E. Hinton: Visualizing High-Dimensional Data Using t-SNE, in: Journal of Machine Learning Research, Vol. 9, (2008), pp. 2579-2605. 3. Aktuální články z profilujících světových časopisů, např.: Neurocomputing, Neural Networks, IEEE Transactions on Neural Networks, etc. |