Guiding Visual Attention in Deep Convolutional Neural Networks Based on Human Eye Movements
This work addresses the challenge of enhancing biological plausibility in AI models for computational neuroscience, though it is incremental as it builds on prior biologically inspired approaches without achieving new human-like gains.
The study investigated whether modifying training examples with human eye tracking data could guide deep convolutional neural networks' visual attention during object recognition, finding that non-human-like attention models focused on significantly different image parts than humans, with effects being category-specific and influencing face detection, but no significant increase in human-likeness was achieved.
Deep Convolutional Neural Networks (DCNNs) were originally inspired by principles of biological vision, have evolved into best current computational models of object recognition, and consequently indicate strong architectural and functional parallelism with the ventral visual pathway throughout comparisons with neuroimaging and neural time series data. As recent advances in deep learning seem to decrease this similarity, computational neuroscience is challenged to reverse-engineer the biological plausibility to obtain useful models. While previous studies have shown that biologically inspired architectures are able to amplify the human-likeness of the models, in this study, we investigate a purely data-driven approach. We use human eye tracking data to directly modify training examples and thereby guide the models' visual attention during object recognition in natural images either towards or away from the focus of human fixations. We compare and validate different manipulation types (i.e., standard, human-like, and non-human-like attention) through GradCAM saliency maps against human participant eye tracking data. Our results demonstrate that the proposed guided focus manipulation works as intended in the negative direction and non-human-like models focus on significantly dissimilar image parts compared to humans. The observed effects were highly category-specific, enhanced by animacy and face presence, developed only after feedforward processing was completed, and indicated a strong influence on face detection. With this approach, however, no significantly increased human-likeness was found. Possible applications of overt visual attention in DCNNs and further implications for theories of face detection are discussed.