scholarly journals A Privacy-Masking Learning Algorithm for Online Distributed Optimization over Time-Varying Unbalanced Digraphs

2021 ◽  
Vol 2021 ◽  
pp. 1-12
Author(s):  
Rong Hu ◽  
Binru Zhang
Keyword(s):  
Differential Privacy ◽  
Learning Algorithm ◽  
Distributed Optimization ◽  
Third Party ◽  
Time Varying ◽  
Objective Functions ◽  
Underlying Network ◽  
Local Convex ◽  
General Time ◽  
Privacy Level

This paper investigates a constrained distributed optimization problem enabled by differential privacy where the underlying network is time-changing with unbalanced digraphs. To solve such a problem, we first propose a differentially private online distributed algorithm by injecting adaptively adjustable Laplace noises. The proposed algorithm can not only protect the privacy of participants without compromising a trusted third party, but also be implemented on more general time-varying unbalanced digraphs. Under mild conditions, we then show that the proposed algorithm can achieve a sublinear expected bound of regret for general local convex objective functions. The result shows that there is a trade-off between the optimization accuracy and privacy level. Finally, numerical simulations are conducted to validate the efficiency of the proposed algorithm.

Iterative learning algorithm with a quadratic criterion for linear time-varying systems

2002 ◽  
Vol 216 (3) ◽  
pp. 309-316 ◽  
Author(s):  
S N Huang ◽  
K K Tan ◽  
T H Lee
Keyword(s):  
Learning Algorithm ◽  
Linear Time ◽  
Tracking Error ◽  
Iterative Learning ◽  
Tracking Performance ◽  
Time Varying ◽  
Control Scheme ◽  
Time Varying Systems ◽  
Simulation Results ◽  
Learning Law

A novel iterative learning controller for linear time-varying systems is developed. The learning law is derived on the basis of a quadratic criterion. This control scheme does not include package information. The advantage of the proposed learning law is that the convergence is guaranteed without the need for empirical choice of parameters. Furthermore, the tracking error on the final iteration will be a class K function of the bounds on the uncertainties. Finally, simulation results reveal that the proposed control has a good setpoint tracking performance.

scholarly journals Manipulation Attacks in Local Differential Privacy

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Albert Cheu ◽  
Adam Smith ◽  
Jonathan Ullman
Keyword(s):  
Systematic Study ◽  
Differential Privacy ◽  
Domain Size ◽  
User Data ◽  
Large Systems ◽  
Cryptographic Techniques ◽  
Fundamental Limitation ◽  
Theoretical Results ◽  
Different Levels ◽  
Privacy Level

Local differential privacy is a widely studied restriction on distributed algorithms that collect aggregates about sensitive user data, and is now deployed in several large systems. We initiate a systematic study of a fundamental limitation of locally differentially private protocols: they are highly vulnerable to adversarial manipulation. While any algorithm can be manipulated by adversaries who lie about their inputs, we show that any noninteractive locally differentially private protocol can be manipulated to a much greater extent---when the privacy level is high, or the domain size is large, a small fraction of users in the protocol can completely obscure the distribution of the honest users' input. We also construct protocols that are optimally robust to manipulation for a variety of common tasks in local differential privacy. Finally, we give simple experiments validating our  theoretical results, and demonstrating that protocols that are optimal without manipulation can have dramatically different levels of robustness to manipulation. Our results suggest caution when deploying local differential privacy and reinforce the importance of efficient cryptographic  techniques for the distributed emulation of centrally differentially private mechanisms.

Kinematic Multi-Objective Optimization of Circular-Toothed Gerotor Pumps by Genetic Algorithm

2017 ◽  
Author(s):  
Andrew J. Robison ◽  
Andrea Vacca
Keyword(s):  
Genetic Algorithm ◽  
Design Method ◽  
Third Party ◽  
Low Flow ◽  
Objective Functions ◽  
Multi Objective Optimization ◽  
Nsga Ii ◽  
Generation Algorithm ◽  
Multi Objective ◽  
Pareto Curve

A gerotor gear generation algorithm has been developed that evaluates key performance objective functions to be minimized or maximized, and then an optimization algorithm is applied to determine the best design. Because of their popularity, circular-toothed gerotors are the focus of this study, and future work can extend this procedure to other gear forms. Parametric equations defining the circular-toothed gear set have been derived and implemented. Two objective functions were used in this kinematic optimization: maximize the ratio of displacement to pump radius, which is a measure of compactness, and minimize the kinematic flow ripple, which can have a negative effect on system dynamics and could be a major source of noise. Designs were constrained to ensure drivability, so the need for additional synchronization gearing is eliminated. The NSGA-II genetic algorithm was then applied to the gear generation algorithm in modeFRONTIER, a commercial software that integrates multi-objective optimization with third-party engineering software. A clear Pareto front was identified, and a multi-criteria decision-making genetic algorithm was used to select three optimal designs with varying priorities of compactness vs low flow variation. In addition, three pumps used in industry were scaled and evaluated with the gear generation algorithm for comparison. The scaled industry pumps were all close to the Pareto curve, but the optimized designs offer a slight kinematic advantage, which demonstrates the usefulness of the proposed gerotor design method.

Sign in / Sign up

Export Citation Format

Share Document