Protecting Privacy through Distributed Computation in Multi-agent Decision Making

Abstract

As large-scale theft of data from corporate servers is becoming increasingly common, it becomes interesting to examine alternatives to the paradigm of centralizing sensitive data into large databases. Instead, one could use cryptography and distributed computation so that sensitive data can be supplied and processed in encrypted form, and only the final result is made known. In this paper, we examine how such a paradigm can be used to implement constraint satisfaction, a technique that can solve a broad class of AI problems such as resource allocation, planning, scheduling, and diagnosis. Most previous work on privacy in constraint satisfaction only attempted to protect specific types of information, in particular the feasibility of particular combinations of decisions. We formalize and extend these restricted notions of privacy by introducing four types of private information, including the feasibility of decisions and the final decisions made, but also the identities of the participants and the topology of the problem. We present distributed algorithms that allow computing solutions to constraint satisfaction problems while maintaining these four types of privacy. We formally prove the privacy properties of these algorithms, and show experiments that compare their respective performance on benchmark problems.

Opponent-Aware Planning with Admissible Privacy Preserving for UGV Security Patrol under Contested Environment

“…Privacy and Security: Many privacy models have been adopted in multi-agent planning according to three different criteria: the information model (imposed privacy [19], induced privacy [20]), the information-sharing scheme (MA-STRIPS [19], subset privacy [21]), and practical privacy guarantees (no privacy [22], weak privacy [23], object cardinality privacy [24], and strong privacy [9]). Privacy can be divided into different categories, such as agent privacy, model privacy, decision privacy, topology privacy, and constraint privacy [4,14]. Here we introduce some widely used types of privacy.…”

“…Privacy-preserving plans represent plans that do not actively disclose sensitive private information. In fact, privacy preservation is the goal pursued by multi-agent planning, which has been a crucial concern for multi-agent systems in some contexts, such as agent negotiation [12], multi-agent reinforcement learning and policy iteration [4,5], deep learning [13], and distributed constraint optimization problems (DCOPs) [14][15][16]. Multi-agent planning (MAP) in cooperative environments aims at generating a sequence of actions to fulfill some specified goals [17].…”

Goal Recognition Control under Network Interdiction Using a Privacy Information Metric

“…Many recent pieces of work explore deception [23] and privacy [24]. For example, deception was investigated in two classes, simulation and dissimulation [23], privacy was investigated in five classes, agent privacy, model privacy, decision privacy, topology privacy, and constraint privacy [25,26]. In cooperative and adversarial environment, a goal-driven intelligent agent would manipulate the goals in the goal-pursuing process, such as sharing or revealing the goals among teammates and meanwhile hide the goals against adversaries.…”

Multiagent Decision Making in Collaborative Decision Networks by Utility Cluster Based Partial Evaluation

“…[1][2][3][4][5][6][7] We consider multiagent, cooperative, simultaneous decision making in partially observable and stochastic environments. Here, simultaneous signifies that each agent decides on actions over multiple decision variables at once.…”