Federated learning is a method of distributed machine learning based on privacy protection, which is characterized by a large number of heterogeneous clients, and often these clients have great differences in communication and computation capabilities, such as servers and smartphones. In this work, we propose a new federated learning algorithm named AAFL, which is based on asynchronous communication and can adaptively assign training models based on the capabilities of different clients to solve the communication efficiency and privacy problems caused by a large number of heterogeneous clients. Our proposed solution can assign training models with different computational complexity according to clients with different capabilities, and reduce the problem that some clients with poor capabilities slow down the training speed of the global model by means of synchronous communication. AAFL can allocate sub-models of different sizes and still aggregate a single global model. The experimental results show that assigning models of different scales according to the capabilities of different clients and using asynchronous communication can improve the training speed and accuracy of the global model to a certain extent, and has a good effect on protecting the privacy of user data.
Federated Learning (FL) is a special distributed machine learning environment. It is jointly trained by many clients under the coordination of a central server. And differential privacy can provide privacy guarantee for FL. While, federated learning, compared with centralized learning, converges at slower speed. And differential privacy exacerbates this trend. In this paper, we propose a novel FL algorithm named DP-FedADMM to solve these problems. We merge differential privacy (DP) into the FedADMM algorithm and propose a method to handle noisy gradients. Under the guidance of the help of the latest results in differential privacy theory, we provide a privacy proof of DP-FedADMM. Through extensive experiments on datasets, it is demonstrated that DP-FedADMM outperforms the currently popular DP-FedAvg algorithm in terms of model convergence speed.
The phenomenon of traffic congestion is now widespread in the traffic of major and medium-sized cities. This study creates a traffic flow diversion system with a backend command, control system management, and a data monitoring visualization large screen surveillance side to enhance operation management, incident response effectiveness, and traffic system. The system leverages relevant expertise in machine learning and visual analytics for target tracking, violation detection, and other tasks. Jetson NX receives deep learning models from Deepstream [2] to enable secure two-way communication between the edge and the cloud. The system also uses signal lights to implement intelligent traffic guidance. In this study, we illustrate the simulation using Pygame, and the simulation outcomes show the efficacy of scheduling.
Visually impaired person cannot perform their activities like ordinary person. Therefore, it becomes more difficult when the visually impaired people working with the computers in their daily life and study. In order to overcome this difficulty, an electronic braille system based on weighted Viterbi algorithm is proposed in this paper. The proposed electronic Braille system including braille input module, speech translation module and text translation module. Along with this system, the electronic braille input, two-way conversion between Braille and Chinese characters, and speech conversion can be realized. The hidden Markov model (HMM) is used to design and implement the transformation, and a weighted Viterbi algorithm is used to solve the model design details and methods. The results presented in this paper show a proper weight on the information provided by static parameters can substantially reduce the error rate, and improve color space conversion accuracy. Finally, it is shown that the system developed in this paper can help visually impaired people for their input without speech assistance, and it can help the visually impaired to study and work better.
As a defence to the SAT-based de-camouflaging attack, a brand new camouflaging strategy (called CamoPerturb) has been proposed. This camouflaging strategy achieves good resistance against the SAT-based attack. However, there is no stable generation algorithm for the perturbation circuit of the CamoPerturb. Furthermore, there is no way to control three key logic: specific output, specific gate and specific minterm, which makes it difficult for the technology to be used in practical circuits. In this paper, a perturbation circuit generation algorithm is proposed. By dividing a limited region, introducing the original input signal and narrowing the minterm range affected by the logic gate, we can find a single minterm perturbation quickly. Furthermore, we can design key logic arbitrarily. The experimental results in ISCAS’ 89 benchmark circuits and the controllers of the OpenSPARC microprocessor show that our method can be applied to the actual circuit stably with very little time consumption. Finally, our method is experimentally analyzed on an IP (intellectual property) core circuit (full adder circuit).
