Print Email Facebook Twitter Understanding weight-magnitude hyperparameters in training binary networks Title Understanding weight-magnitude hyperparameters in training binary networks Author Quist, Joris (TU Delft Electrical Engineering, Mathematics and Computer Science) Contributor Li, Y. (mentor) van Gemert, J.C. (mentor) Lofi, C. (graduation committee) Degree granting institution Delft University of Technology Programme Computer Science | Data Science and Technology Date 2022-11-07 Abstract Binary Neural Networks (BNNs) are compact and efficient by using binary weights instead of real-valued weights. Current BNNs use latent real-valued weights during training, where several training hyper-parameters are inherited from real-valued networks. The interpretation of several of these hyperparameters is based on the magnitude of the real-valued weights. For BNNs, however, the magnitude of binary weights is not meaningful, and thus it is unclear what these hyperparameters actually do. One example is weight-decay, which aims to keep the magnitude of real-valued weights small. Other examples are latent weight initialization, the learning rate, and learning rate decay, which influence the magnitude of the real-valued weights. The magnitude is interpretable for real-valued weights, but loses its meaning for binary weights. In this paper we offer a new interpretation of these magnitude-based hyperparameters based on higher-order gradient filtering during network optimization. Our analysis makes it possible to understand how magnitude-based hyperparameters influence the training of binary networks which allows for new optimization filters specifically designed for binary neural networks that are independent of their real-valued interpretation. Moreover, our improved understanding reduces the number of hyperparameters, which in turn eases the hyperparameter tuning effort which may lead to better hyperparameter values for improved accuracy. Subject Deep LearningBinary Neural NetworksHyperparameter Optimization To reference this document use: http://resolver.tudelft.nl/uuid:fe011a49-787e-49cd-a5b3-2b63a8c8c4ef Part of collection Student theses Document type master thesis Rights © 2022 Joris Quist Files PDF master_thesis_joris_quist.pdf 965.42 KB Close viewer /islandora/object/uuid:fe011a49-787e-49cd-a5b3-2b63a8c8c4ef/datastream/OBJ/view