Adobe PDF (1.43 MB)
Title Details:
Optimizations for Deep Model Training
Authors: Likothanassis, Spiridon
Koutsomitropoulos, Dimitrios
Description:
Abstract:
This chapter presents some basic techniques for training deep neural networks and in particular, the selection of an appropriate cost function, called cross-entropy, four normalization methods (L1, L2, dropout and expansion of the training set), methods for initializing network weights, and heuristics for selecting the optimal network hyperparameters.
Linguistic Editors: Kanari, Vasiliki
Graphic Editors: Moustani, Evagelia
Type: Chapter
Creation Date: 09-11-2023
Item Details:
License: Attribution - NonCommercial - ShareAlike 4.0 International (CC BY-NC-SA 4.0)
Handle http://hdl.handle.net/11419/11355
Bibliographic Reference: Likothanassis, S., & Koutsomitropoulos, D. (2023). Optimizations for Deep Model Training [Chapter]. In Likothanassis, S., & Koutsomitropoulos, D. 2023. Computational intelligence and deep learning [Undergraduate textbook]. Kallipos, Open Academic Editions. https://hdl.handle.net/11419/11355
Language: Greek
Is Part of: Computational intelligence and deep learning
Publication Origin: Kallipos, Open Academic Editions