HSI: Human Saliency Imitator for Benchmarking Saliency-Based Model Explanations


  • Yi Yang Huawei Research Hong Kong
  • Yueyuan Zheng Huawei Research Hong Kong Department of Psychology, University of Hong Kong
  • Didan Deng Huawei Research Hong Kong Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology
  • Jindi Zhang Huawei Research Hong Kong
  • Yongxiang Huang Huawei Research Hong Kong
  • Yumeng Yang Department of Psychology, University of Hong Kong
  • Janet H. Hsiao Department of Psychology, University of Hong Kong State Key Laboratory of Brain and Cognitive Sciences, University of Hong Kong Institute of Data Science, University of Hong Kong
  • Caleb Chen Cao Huawei Research Hong Kong




Explainable AI, XAI Evaluation, Benchmarking, Human Saliency, Crowdsourcing


Model explanations are generated by XAI (explainable AI) methods to help people understand and interpret machine learning models. To study XAI methods from the human perspective, we propose a human-based benchmark dataset, i.e., human saliency benchmark (HSB), for evaluating saliency-based XAI methods. Different from existing human saliency annotations where class-related features are manually and subjectively labeled, this benchmark collects more objective human attention on vision information with a precise eye-tracking device and a novel crowdsourcing experiment. Taking the labor cost of human experiment into consideration, we further explore the potential of utilizing a prediction model trained on HSB to mimic saliency annotating by humans. Hence, a dense prediction problem is formulated, and we propose an encoder-decoder architecture which combines multi-modal and multi-scale features to produce the human saliency maps. Accordingly, a pretraining-finetuning method is designed to address the model training problem. Finally, we arrive at a model trained on HSB named human saliency imitator (HSI). We show, through an extensive evaluation, that HSI can successfully predict human saliency on our HSB dataset, and the HSI-generated human saliency dataset on ImageNet showcases the ability of benchmarking XAI methods both qualitatively and quantitatively.




How to Cite

Yang, Y., Zheng, Y., Deng, D., Zhang, J., Huang, Y., Yang, Y., Hsiao, J. H., & Cao, C. C. (2022). HSI: Human Saliency Imitator for Benchmarking Saliency-Based Model Explanations. Proceedings of the AAAI Conference on Human Computation and Crowdsourcing, 10(1), 231-242. https://doi.org/10.1609/hcomp.v10i1.22002