Stochastic gradient descent (SGD) is a widely used optimization algorithm in machine learning. In order to accelerate the convergence of SGD, a few advanced techniques have been developed in recent years, including variance reduction, stochastic coordinate sampling, and Nesterov’s acceleration method. Furthermore, in order to improve the training speed and/or leverage larger-scale training data...

This paper extends to pomsets without auto-concurrency the fundamental notion of asynchronous cellular automata (ACA) which was originally introduced for traces by Zielonka. We generalize to pomsets the notion of asynchronous mapping introduced by Cori, Métivier and Zielonka and we show how to construct a deterministic ACA from an asynchronous mapping. Then we investigate the relation between t...

We present a new algorithm to compute the watershed transform on a massively parallel machine. Logic minimization, together with a novel approach to the notion of dynamics, allow the implementation of the watershed algorithm on a cellular automata grid, such as the programmable retina. We then investigate the profit acquired from computing some parts of the algorithm in an asynchronous way, in ...

The interactive computation paradigm is reviewed and a particular example is extended to form the stochastic analog of a computational process via a transcription of a minimal Turing Machine into an equivalent asynchronous Cellular Automaton with an exponential waiting times distribution of effective transitions. Furthermore, a special toolbox for analytic derivation of recursive relations of i...