Search

A variety of modern AI products essentially require raw user data for training diverse machine learning models. With the increasing concern on data privacy, federated learning, a decentralized learning framework, enables privacy-preserving training of models by iteratively aggregating model updates from participants, instead of aggregating raw data. Since all the participants, i.e., mobile devices, need to transfer their local model updates concurrently and iteratively over mobile edge networks, the network is easily overloaded, leading to a high risk of transmission failures. Although previous works on transmission protocols have already tried their best to avoid transmission collisions, the number of iterative concurrent transmissions should be fundamentally decreased. Inspired by the fact that raw data are often generated unevenly among devices, those devices with a small proportion of data could be properly excluded since they have little effect on the convergence of models. To further guarantee the accuracy of models, we propose to properly select a subset of devices as participants to ensure the given proportion of involved data. Correspondingly, we propose to minimize the risk against the transmission failures during model updates. Afterwards, we design a randomized algorithm ( $ran$ RFL) to choose suitable participants by using a series of delicately calculated probabilities, and prove that the result is concentrated on its optimum with high probability. Extensive simulations show that through delicate participant selection, $ran$ RFL decreases the maximal error rate of model updates by up to 38.3% compared with the state-of-the-art schemas.

The understanding of circuit diagram is very important for the study of electrical engineering. Existing circuit diagram simulation tools are mostly based on GUI interface and rely on users to click or drag icons with mouse, which requires them to be familiar with the software and distracts a great deal of their attention from the circuit diagram itself. Although a lot of previous works have devoted to designing algorithmic solution to recognize hand-drawn circuit diagrams automatically, there still exists strict constraints on users’ drawing habits and stroke orders. In order to address these inconveniences, this paper proposes a novel sketch recognition algorithm named $LS^{4}D$ . It uses graph to model the input strokes and their relationships, and leverages cycles by local strokes to detect some circuit components. Theoretical derivations have demonstrated that $LS^{4}D$ can efficiently recognize diagrams with different drawing styles and arbitrary stroke orders. To furthermore illustrate the practical value of the proposed approach, we construct a prototype of pen-based circuit diagram system based on $LS^{4}D$ , which enables users to draw circuit diagrams directly on the digital screen without any other restriction. An experiment of 158 samples collected from 17 users is conducted on the designed platform. Our approach has achieved 93.04% recognition accuracy and overall 4.53 from a 5-scale user satisfaction rating, which outperforms previous state-of-the-art methods numerically. It is shown that the same approach can also be generalized to many other sketch recognition applications with minor modifications. To facilitate future researches and applications, we publish our source code, model, and training data at https://github.com/Huage001/ Graph-Based-Circuit-Painter.