Distributed training of Machine Learning models in edge Internet Things (IoT) environments is challenging because three main points. First, resource-constrained devices have large times and limited energy budget. Second, resource heterogeneity IoT slows down the global model due to presence slower (stragglers). Finally, varying operational conditions, such as network bandwidth, computing resour...

Neural sequence labeling (NSL) aims at assigning labels for input language tokens, which covers a broad range of applications, such as named entity recognition (NER) and slot filling, etc. However, the satisfying results achieved by traditional supervised-based approaches heavily depend on large amounts human annotation data, may not be feasible in real-world scenarios due to data privacy compu...

Software development sometimes requires to statically predict the quantity of resources—such as memory and time—that is needed to execute a program. The difficulties of manual resource analysis led to extensive research on automatic methods for quantitative resource analysis. Recently we developed an automatic amortized analysis to compute polynomial resource bounds for first-order functional p...

One of the key challenges for multiagent systems (MAS) is optimizing performance in limited resource environments. Previous research in this area has focused on the problems of 1) resource allocation and arbitration, and 2) bounded rationality, which describe the relationship between resource constraints and both agent reasoning and actuation. However, less work exists addressing the effect of ...

ABSTRACT Integration of context interfaces in mobile application design could be based on different goals, such as provision and exchange of resource information of different type, application adaptation to specific context, efficient interactions between user and application, visibility of specific resource information, building of context aware mobile infrastructures, increased robustness, fa...

Static techniques for deriving upper bounds on the resource consumption of programs have been extensively studied. However, there are applications that require more fine-grained information such as the difference between upper and lower bounds or the guaranty that the resource usage of a program does not differ for certain inputs. This article presents two novel substructural type systems for d...

We introduce and study a functional language R, having two main features. R has the same computational power of the-calculus. R enjoys the resource-awareness of the typed/typable functional languages which encode the Intuitionistic Linear Logic.

Energy efficiency and thermal management have become major concerns in both embedded and HPC systems. The progress of silicon technology and the subsequent growth of the dark silicon phenomena are negatively affecting the reliability of computing systems. As a result, in the next future we expect run-time variability to increase in terms of both performance and computing resources availability....

In recent years, the Cloud Computing paradigm has proven effective in scaling dynamically the number of servers according to simple performance metrics and the incoming workload. However while some applications are able to scale-out, as current scaling metrics do not relate system performance to sales, hosting costs and profits are not optimized completely. The following article proposes a nove...

Resource allocation and straggler mitigation (via “speculative” copies) are two key building blocks for analytics frameworks. Today, the two solutions are largely decoupled from each other, losing the opportunities of joint optimization. Resource allocation across jobs assumes that each job runs a fixed set of tasks, ignoring their need to dynamically run speculative copies for stragglers. Cons...