3 accepted papers
In the process of training Neural Networks, pre-training is an unsupervised training process that uses automatically generated labels for real end-goal task inputs. It usually precedes a supervised training stage, can improve neural network performance, and can reduce training loss. In this work, we…
During training, artificial neural networks might not converge to a global minimum. Usually, using gradient descent, the training procedure cause the network to stroll in the high-dimensional weights’ space. This stroll passes adjacently to local minima and locations in the geometry of loss landscap…
In this work, we propose a method for detection of locations with subjective significance in the visual environment using Information Constrained Control (ICC). ICC is a model that takes into consideration not only the goal but also the complexity of the control needed to achieve it, characterized b…