Showing items 1 - 2 of 2

Your selections:

Adversarial learning-based bias mitigation for fatigue driving detection in fair-intelligent IoV

- Han, Mingzhe, Wu, Jun, Bashir, Ali, Yang, Wu, Imran, Muhammad, Nasser, Nidal

  • Authors: Han, Mingzhe , Wu, Jun , Bashir, Ali , Yang, Wu , Imran, Muhammad , Nasser, Nidal
  • Date: 2020
  • Type: Text , Conference paper
  • Relation: 2020 IEEE Global Communications Conference, GLOBECOM 2020, Virtual, Taipei, China, 7 to 11 December 2020, 2020 IEEE Global Communications Conference, GLOBECOM 2020 - Proceedings
  • Full Text: false
  • Reviewed:
  • Description: Fatigue driving is one of main causes of traffic accidents. To avoid such traffic accidents, divers' fatigue detection has been used in Intelligent Internet of Vehicles (IIoV). IIoV usually dynamically allocate computing resources according to drivers' fatigue degree to improve the real-time of fatigue detection model. However, the traditional fatigue detection model may have bias on certain groups, which would further cause unfair resource allocation. To solve the problem, this paper proposes an improved IIoV framework, named Fair-Intelligent Internet of Vehicles (FIIoV). Compared with IIoV, we improve two layers in FIIoV, i.e., the detection layer and the normalization layer. The detection layer uses Convolutional Neural Network (CNN) to detect drivers' fatigue degree, and then uses adversarial network to achieve fairness of detection models. The normalization layer achieves the distribution of different sensitive feature values from historical detection results generated in the detection layer, and then uses the distribution to normalize the output of the detection layer to improve the fairness and accuracy of fatigue detection models. Simulation results show that both accuracy and fairness of FIIoV is improved compared with the original IIoV. © 2020 IEEE.

SCA-LFD : side-channel analysis-based load forecasting disturbance in the energy internet

- Ding, Li, Wu, Jun, Li, Changlian, Jolfaei, Alireza, Zheng, Xi

  • Authors: Ding, Li , Wu, Jun , Li, Changlian , Jolfaei, Alireza , Zheng, Xi
  • Date: 2023
  • Type: Text , Journal article
  • Relation: IEEE Transactions on Industrial Electronics Vol. 70, no. 3 (2023), p. 3199-3208
  • Full Text: false
  • Reviewed:
  • Description: The energy Internet (EI) equipment may face threats that attackers poison federated learning (FL) models to disturb electricity load forecasting. To mitigate this vulnerability, it is important to study load forecasting disturbance approaches. This article proposes a side-channel analysis (SCA)-based disturbance approach. First, we design an FL SCA scheme to extract power information from the FL chip running forecasting model. Second, we propose an FL data speculation method using an optimized convolutional neural network trained with SCA information. Third, we design a label-flipping-based poisoning scheme with speculated data characteristics for load forecasting disturbance. Experimental results show attackers can successfully poison and disturb FL-based load forecasting. The average accuracy of EI load data speculation is 99.8%. This work is the first to study EI load forecasting disturbance from an SCA perspective. © 1982-2012 IEEE.
  • 1