• Home
  • Search
  • Optimal Attack Strategies Against Predictors - Learning From Expert Advice
  • Cite Icon14
  • https://doi.org/10.1109/tifs.2017.2718488Copy DOI Icon

Optimal Attack Strategies Against Predictors - Learning From Expert Advice

Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Motivated by many real-world examples, such as recommendation systems or sensor fusion, and aiming to capture the influence of malicious experts who intentionally degrade the performance of learning systems, we analyze optimal adversarial strategies against the weighted average prediction algorithm in the learning with expert advice framework. All but one expert is honest and the malicious expert’s goal is to sabotage the performance of the algorithm by strategically providing dishonest recommendations. We formulate the problem as a Markov decision process and analyze it under various settings. For the logarithmic loss, somewhat surprisingly, we prove that the optimal strategy for the adversary is the greedy policy, i.e., lying at every step. For the absolute loss, in the 2-experts, discounted cost setting, we prove that the optimal strategy is a threshold policy, where the malicious expert tells the truth until he earns enough weight and then lies afterwards. We extend the results to the infinite horizon problem and find the exact thresholds for the stationary optimal policy. Finally, we use a mean field approach in the $N$ -experts setting to find the optimal strategy when the predictions of the honest experts are independent and identically distributed. We justify our results using simulations throughout this paper.

Similar Papers
  • Conference Article
  • Citations245

Remote estimation of the Wiener process over a channel with random delay

  • Jun 01, 2017
  • Yin Sun +2
  • Conference Article
  • Citations111

Learning and adaptivity in interactive recommender systems

  • Aug 19, 2007
  • Tariq Mahmood +1
  • Research Article
  • Citations64

When blockchain meets AI: Optimal mining strategy achieved by machine learning

  • Feb 01, 2021
  • International Journal of Intelligent Systems
  • Taotao Wang +2
  • Book Chapter

Fault Diagnosis Strategy Optimization Under Unreliable and Multivalued Tests

  • Oct 02, 2019
  • Yajun Liang +4
  • Research Article
  • Citations25

A Novel Transmission Scheduling Based on Deep Reinforcement Learning in Software-Defined Maritime Communication Networks

  • Dec 01, 2019
  • IEEE Transactions on Cognitive Communications and Networking
  • Tingting Yang +4
  • Book Chapter
  • Citations10

Consensus Algorithm Analysis in Blockchain: PoW and Raft

  • Oct 22, 2021
  • Taotao Wang +2
  • Research Article
  • Citations80

Modeling of Intrusion and Defense for Assessment of Cyber Security at Power Substations

  • Jul 01, 2018
  • IEEE Transactions on Smart Grid
  • Ying Chen +2
  • Conference Article

Charging Scheduling of Enterprise Private Parking-lot with Renewable Power: A simulation method based on Markov Decision Processes

  • Jul 01, 2019
  • Fangzhu Ming +3
  • Research Article
  • Citations3

Application of Markov decision processes to search problems

  • Jul 01, 1995
  • Decision Support Systems
  • Leo B Hartman +1
  • Research Article
  • Citations1

Approximation of Discounted Minimax Markov Control Problems and Zero-Sum Markov Games Using Hausdorff and Wasserstein Distances

  • Mar 19, 2018
  • Dynamic Games and Applications
  • François Dufour +1
  • Research Article
  • Citations4

Crosslayer parameter configuration for TCP throughput improvement in cognitive radio networks

  • Jan 01, 2013
  • Acta Physica Sinica
  • Jiang Hong +2
  • Conference Article
  • Citations8

Optimality results for a simple flow control problem

  • Dec 01, 1987
  • Dye-Jyun Ma +1
  • Research Article
  • Citations5

Optimal investment strategy on data analytics capabilities of startups via Markov decision analysis

  • Mar 01, 2024
  • Decision Analytics Journal
  • Maarten Voorneveld +1
  • Research Article
  • Citations2

A Markov Decision Process (MDP) based Load Balancing Algorithm for Multi-cell Networks with Multi-carriers

  • Oct 31, 2014
  • KSII Transactions on Internet and Information Systems
  • Janghoon Yang
  • Research Article
  • Citations5

Existence of an Optimal Stationary Policy in a Markov Decision Process

  • Jan 01, 1965
  • Theory of Probability & Its Applications
  • I V Romanovskii
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.