Explaining image enhancement black-box methods through a path planning based algorithm

被引:1
|
作者
Cotogni, Marco [1 ]
Cusano, Claudio [1 ]
机构
[1] Univ Pavia, Dept Elect Comp & Biomed Engn, Via Ferrata 1, I-27100 Pavia, Italy
关键词
Explainable AI; Image enhancement; Heuristic search; Path planning;
D O I
10.1007/s11042-023-15648-8
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Nowadays, image-to-image translation methods, are the state of the art for the enhancement of natural images. Even if they usually show high performance in terms of accuracy, they often suffer from several limitations such as the generation of artifacts and the scalability to high resolutions. Moreover, their main drawback is the completely black-box approach that does not allow to provide the final user with any insight about the enhancement processes applied. In this paper we present a path planning algorithm which provides a step-by-step explanation of the output produced by state of the art enhancement methods. This algorithm, called eXIE, uses a variant of A(*) to emulate the enhancement process of another method through the application of an equivalent sequence of enhancing operators. We applied eXIE to explain the output of several state-of-the-art models trained on the Five-K dataset, obtaining sequences of enhancing operators able to produce very similar results in terms of performance and overcoming the huge limitation of poor interpretability of the best performing algorithms.
引用
收藏
页码:8043 / 8062
页数:20
相关论文
共 50 条
  • [21] Execution Path Detection through Dynamic Analysis in Black-Box Testing Environments
    Whitworth, Frank
    PROCEEDINGS OF THE 37TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING, ASE 2022, 2022,
  • [22] Black-box modeling of residential HVAC system and comparison of gray-box and black-box modeling methods
    Afram, Abdul
    Janabi-Sharifi, Farrokh
    ENERGY AND BUILDINGS, 2015, 94 : 121 - 149
  • [23] Exploiting auto-encoders for explaining black-box classifiers
    Guidotti, Riccardo
    INTELLIGENZA ARTIFICIALE, 2022, 16 (01) : 115 - 129
  • [24] A General Method for Visualizing and Explaining Black-Box Regression Models
    Strumbelj, Erik
    Kononenko, Igor
    ADAPTIVE AND NATURAL COMPUTING ALGORITHMS, PT II, 2011, 6594 : 21 - 30
  • [25] Explaining Black-Box Algorithms Using Probabilistic Contrastive Counterfactuals
    Galhotra, Sainyam
    Pradhan, Romila
    Salimi, Babak
    SIGMOD '21: PROCEEDINGS OF THE 2021 INTERNATIONAL CONFERENCE ON MANAGEMENT OF DATA, 2021, : 577 - 590
  • [26] Analyzing and Explaining Black-Box Models for Online Malware Detection
    Manthena, Harikha
    Kimmel, Jeffrey C.
    Abdelsalam, Mahmoud
    Gupta, Maanak
    IEEE ACCESS, 2023, 11 : 25237 - 25252
  • [27] Explaining Black-box Predictions by Generating Local Meaningful Perturbations
    Verma, Tejaswani
    Lingenfelder, Christoph
    Klakow, Dietrich
    INTERNATIONAL JOURNAL OF SEMANTIC COMPUTING, 2022, 16 (01) : 47 - 68
  • [28] Generating Causal Hypotheses for Explaining Black-Box Industrial Processes
    Balzereit, Kaja
    Diedrich, Alexander
    Kubus, Daniel
    Ginster, Jonas
    Bunte, Andreas
    2022 IEEE 5TH INTERNATIONAL CONFERENCE ON INDUSTRIAL CYBER-PHYSICAL SYSTEMS, ICPS, 2022,
  • [29] Composition of relational features with an application to explaining black-box predictors
    Ashwin Srinivasan
    A. Baskar
    Tirtharaj Dash
    Devanshu Shah
    Machine Learning, 2024, 113 : 1091 - 1132
  • [30] Composition of relational features with an application to explaining black-box predictors
    Srinivasan, Ashwin
    Baskar, A.
    Dash, Tirtharaj
    Shah, Devanshu
    MACHINE LEARNING, 2024, 113 (03) : 1091 - 1132