Machine learning over fully distributed data poses an important problem in peer-to-peer (P2P) applications. In this model we have one data record at each network node, but without the possibility to move raw data due to privacy considerations. For example, user profiles, ratings, history, or sensor readings can represent this case. This problem is difficult, because there is no possibility to l...

We address the issue of speeding up the training of convolutional neural networks by studying a distributed method adapted to stochastic gradient descent. Our parallel optimization setup uses several threads, each applying individual gradient descents on a local variable. We propose a new way of sharing information between different threads based on gossip algorithms that show good consensus co...

Problems of cooperation are frequent among living organisms, but they difficult to solve. Humans have been able produce large-scale unrelated individuals through reputation systems. A challenging puzzle, however, is how can guide behavior if in most cases it not shared publicly and assigned others privately. We confirm that obtain agents playing the Prisoner’s Dilemma when reputations individua...

This paper gives drastically faster gossip algorithms to compute exact and approximate quantiles. Gossip algorithms, which allow each node to contact a uniformly random other node in each round, have been intensely studied and been adopted in many applications due to their fast convergence and their robustness to failures. Kempe et al. [KDG03, FOCS’03] gave gossip algorithms to compute importan...

During recent years, there has been a rapid growth in deployment of gossip-based protocol in many multicast applications. In a typical gossip-based protocol, each node acts as dual roles of receiver and sender, independently exchanging data with its neighbors to facilitate scalability and resilience. However, most of previous work in this literature seldom considered cheating issue of end users...

Gossip-based broadcast algorithms have been considered as a viable alternative to traditional deterministic reliable broadcast algorithms in large scale environments. However, these algorithms focus on broadcasting events inside a large group of processes, while the multicasting of events to a subset of processes in a group only, potentially varying for every event, has not been considered. We ...

Gossip is a well-known technique for distributed computing in an arbitrarily connected network, that can be adopted effectively in wireless sensor networks. Gossip algorithms have been widely studied in previous literature, but mostly from a theoretical point of view. The aim of this paper is to verify the behavior of the gossip approach in practical scenarios, through the analysis and interpre...

Gossip-based protocols are simple, robust and scalable and have been consistently applied in many distributed, mostly wired, settings. Most validation in this area has been so far empirical and there is a clear lack of a theoretical counterpart clearly defining what can and cannot be computed with gossip-based protocols. Population protocols, on the other hand, provide a clear and formal model ...