Adversarial Attacks Assessment of Salient Object Detection via Symbolic Learning
Authors:
Gustavo Olague,
Roberto Pineda,
Gerardo Ibarra-Vazquez,
Matthieu Olague,
Axel Martinez,
Sambit Bakshi,
Jonathan Vargas,
Isnardo Reducindo
Abstract:
Machine learning is at the center of mainstream technology and outperforms classical approaches to handcrafted feature design. Aside from its learning process for artificial feature extraction, it has an end-to-end paradigm from input to output, reaching outstandingly accurate results. However, security concerns about its robustness to malicious and imperceptible perturbations have drawn attention…
▽ More
Machine learning is at the center of mainstream technology and outperforms classical approaches to handcrafted feature design. Aside from its learning process for artificial feature extraction, it has an end-to-end paradigm from input to output, reaching outstandingly accurate results. However, security concerns about its robustness to malicious and imperceptible perturbations have drawn attention since its prediction can be changed entirely. Salient object detection is a research area where deep convolutional neural networks have proven effective but whose trustworthiness represents a significant issue requiring analysis and solutions to hackers' attacks. Brain programming is a kind of symbolic learning in the vein of good old-fashioned artificial intelligence. This work provides evidence that symbolic learning robustness is crucial in designing reliable visual attention systems since it can withstand even the most intense perturbations. We test this evolutionary computation methodology against several adversarial attacks and noise perturbations using standard databases and a real-world problem of a shorebird called the Snowy Plover portraying a visual attention task. We compare our methodology with five different deep learning approaches, proving that they do not match the symbolic paradigm regarding robustness. All neural networks suffer significant performance losses, while brain programming stands its ground and remains unaffected. Also, by studying the Snowy Plover, we remark on the importance of security in surveillance activities regarding wildlife protection and conservation.
△ Less
Submitted 11 September, 2023;
originally announced September 2023.
