In practice, data augmentation is assigned a predefined budget in terms of newly created samples per epoch. When using several types of data augmentation, the budget is usually uniformly distributed over the set of augmentations but one can wonder if this budget should not be allocated to each type in a more efficient way. This paper leverages online learning to allocate on the fly this budget as part of neural network training. This meta-algorithm can be run at almost no extra cost as it exploits gradient based signals to determine which type of data augmentation should be preferred. Experiments suggest that this strategy can save computation time and thus goes in the way of greener machine learning practices.

Significant cell-to-cell heterogeneity is ubiquitously observed in isogenic cell populations. Consequently, parameters of models of intracellular processes, usually fitted to population-averaged data, should rather be fitted to individual cells to obtain a population of models of similar but non-identical individuals. Here, we propose a quantitative modeling framework that attributes specific parameter values to single cells for a standard model of gene expression. We combine high quality single-cell measurements of the response of yeast cells to repeated hyperosmotic shocks and state-of-the-art statistical inference approaches for mixed-effects models to infer multidimensional parameter distributions describing the population, and then derive specific parameters for individual cells. The analysis of single-cell parameters shows that single-cell identity (e.g. gene expression dynamics, cell size, growth rate, mother-daughter relationships) is, at least partially, captured by the parameter values of gene expression models (e.g. rates of transcription, translation and degradation). Our approach shows how to use the rich information contained into longitudinal single-cell data to infer parameters that can faithfully represent single-cell identity.

A 2 CCS is a conservative extension of CCS, enriched with an operator of strong prefixing, enabling the modeling of atomic sequences and multi-party synchronization (realized as an atomic sequence of binary synchronizations); the classic dining philosophers problem is used to illustrate the approach. A step semantics for A 2 CCS is also presented directly as a labeled transition system. A safe Petri net semantics for this language is presented, following the approach of Degano, De Nicola, Montanari and Olderog. We prove that a process p and its associated net Net(p) are interleaving bisimilar (Theorem 5.1). Moreover, to support the claim that the intended concurrency is well-represented in the net, we also prove that a process p and its associated net Net(p) are step bisimilar (Theorem 5.2).

Priority is a frequently used feature of many computational systems. In this paper we study the expressiveness of two process algebras enriched with different priority mechanisms. In particular, we consider a finite (that is, recursion-free) fragment of asynchronous CCS with global priority (FAP, for short) and Phillips' CPG (CCS with local priority), and contrast their expressive power with that of two non-prioritised calculi, namely the π-calculus and its broadcast-based version, called bπ. We prove, by means of leader-election-based separation results, that, under certain conditions, there exists no encoding of FAP in π-Calculus or CPG. Moreover, we single out another problem in distributed computing, which we call thelast man standingproblem (LMS for short), that better reveals the gap between the two prioritised calculi above and the two non-prioritised ones, by proving that there exists no parallel-preserving encoding of the prioritised calculi in the non-prioritised calculi retaining anysincere(complete but partially correct, that is, admitting divergence or premature termination) semantics.