{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,15]],"date-time":"2026-01-15T09:24:29Z","timestamp":1768469069451,"version":"3.49.0"},"reference-count":92,"publisher":"MIT Press","issue":"9","content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2024,8,19]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Humans actively observe the visual surroundings by focusing on salient objects and ignoring trivial details. However, computer vision models based on convolutional neural networks (CNN) often analyze visual input all at once through a single feedforward pass. In this study, we designed a dual-stream vision model inspired by the human brain. This model features retina-like input layers and includes two streams: one determining the next point of focus (the fixation), while the other interprets the visuals surrounding the fixation. Trained on image recognition, this model examines an image through a sequence of fixations, each time focusing on different parts, thereby progressively building a representation of the image. We evaluated this model against various benchmarks in terms of object recognition, gaze behavior, and adversarial robustness. Our findings suggest that the model can attend and gaze in ways similar to humans without being explicitly trained to mimic human attention and that the model can enhance robustness against adversarial attacks due to its retinal sampling and recurrent processing. In particular, the model can correct its perceptual errors by taking more glances, setting itself apart from all feedforward-only models. In conclusion, the interactions of retinal sampling, eye movement, and recurrent dynamics are important to human-like visual exploration and inference.<\/jats:p>","DOI":"10.1162\/neco_a_01688","type":"journal-article","created":{"date-parts":[[2024,8,6]],"date-time":"2024-08-06T20:28:53Z","timestamp":1722976133000},"page":"1713-1743","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":2,"title":["Human Eyes\u2013Inspired Recurrent Neural Networks Are More Robust Against Adversarial Noises"],"prefix":"10.1162","volume":"36","author":[{"given":"Minkyu","family":"Choi","sequence":"first","affiliation":[{"name":"Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI 48105, U.S.A. cminkyu@umich.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yizhen","family":"Zhang","sequence":"additional","affiliation":[{"name":"Department of Neurological Surgery, University of California, San Francisco, CA 94143, U.S.A. yizhen.zhang@ucsf.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Kuan","family":"Han","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI 48105, U.S.A. kuanhan@umich.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xiaokai","family":"Wang","sequence":"additional","affiliation":[{"name":"Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI 48109, U.S.A. xiaokaiw@umich.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhongming","family":"Liu","sequence":"additional","affiliation":[{"name":"Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI 48109, U.S.A."},{"name":"Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI 48105, U.S.A. zmliu@umich.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"281","published-online":{"date-parts":[[2024,8,19]]},"reference":[{"issue":"10","key":"2024082018250840100_bib1","doi-asserted-by":"publisher","first-page":"e1005743","DOI":"10.1371\/journal.pcbi.1005743","article-title":"Object detection through search with a foveated visual system","volume":"13","author":"Akbas","year":"2017","journal-title":"PLOS Computational Biology"},{"key":"2024082018250840100_bib2","first-page":"284","article-title":"Synthesizing robust adversarial examples","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Athalye","year":"2018"},{"issue":"12","key":"2024082018250840100_bib3","doi-asserted-by":"crossref","DOI":"10.1167\/9.12.13","article-title":"A summary-statistic representation in peripheral vision explains visual crowding","volume":"9","author":"Balas","year":"2009","journal-title":"Journal of Vision"},{"key":"2024082018250840100_bib4","doi-asserted-by":"publisher","first-page":"154","DOI":"10.1016\/j.neucom.2020.03.060","article-title":"Human scanpath prediction based on deep convolutional saccadic model","volume":"404","author":"Bao","year":"2020","journal-title":"Neurocomputing"},{"issue":"6439","key":"2024082018250840100_bib5","doi-asserted-by":"publisher","DOI":"10.1126\/science.aav9436","article-title":"Neural population control via deep image synthesis","volume":"364","author":"Bashivan","year":"2019","journal-title":"Science"},{"key":"2024082018250840100_bib6","doi-asserted-by":"publisher","first-page":"270","DOI":"10.1016\/j.neucom.2020.07.047","article-title":"Modeling bottom-up and top-down attention with a neurodynamic model of V1","volume":"417","author":"Berga","year":"2020","journal-title":"Neurocomputing"},{"key":"2024082018250840100_bib7","author":"Berrios","year":"2022","journal-title":"Joint rotational invariance and adversarial training of a dual-stream transformer yields state of the art brain-score for area V4."},{"issue":"5241","key":"2024082018250840100_bib8","doi-asserted-by":"publisher","first-page":"177","DOI":"10.1038\/226177a0","article-title":"Interaction effects in parafoveal letter recognition","volume":"226","author":"Bouma","year":"1970","journal-title":"Nature"},{"issue":"45","key":"2024082018250840100_bib9","doi-asserted-by":"publisher","first-page":"19525","DOI":"10.1073\/pnas.1006076107","article-title":"Bayesian model of dynamic image stabilization in the visual system","volume":"107","author":"Burak","year":"2010","journal-title":"Proceedings of the National Academy of Sciences"},{"key":"2024082018250840100_bib10","doi-asserted-by":"crossref","first-page":"39","DOI":"10.1109\/SP.2017.49","article-title":"Towards evaluating the robustness of neural networks","volume-title":"Proceedings of the 2017 IEEE Symposium on Security and Privacy","author":"Carlini","year":"2017"},{"key":"2024082018250840100_bib11","author":"Cheung","year":"2016","journal-title":"Emergence of foveal image sampling from learning to attend in visual scenes"},{"key":"2024082018250840100_bib12","author":"Choi","year":"2023","journal-title":"A dual-stream neural network explains the functional segregation of dorsal and ventral visual pathways in human brains"},{"key":"2024082018250840100_bib13","author":"Chung","year":"2014","journal-title":"Empirical evaluation of gated recurrent neural networks on sequence modeling"},{"issue":"1","key":"2024082018250840100_bib14","doi-asserted-by":"publisher","first-page":"319","DOI":"10.1146\/annurev.neuro.22.1.319","article-title":"Space and attention in parietal cortex","volume":"22","author":"Colby","year":"1999","journal-title":"Annual Review of Neuroscience"},{"issue":"8","key":"2024082018250840100_bib15","doi-asserted-by":"publisher","first-page":"1503","DOI":"10.1364\/JOSAA.4.001503","article-title":"Psychophysical estimate of extrafoveal cone spacing","volume":"4","author":"Coletta","year":"1987","journal-title":"JOSA A"},{"issue":"4","key":"2024082018250840100_bib16","doi-asserted-by":"publisher","first-page":"544","DOI":"10.1002\/cne.902260408","article-title":"The representation of the visual field in parvicellular and magnocellular layers of the lateral geniculate nucleus in the macaque monkey","volume":"226","author":"Connolly","year":"1984","journal-title":"Journal of Comparative Neurology"},{"issue":"3","key":"2024082018250840100_bib17","doi-asserted-by":"publisher","first-page":"201","DOI":"10.1038\/nrn755","article-title":"Control of goal-directed and stimulus-driven attention in the brain","volume":"3","author":"Corbetta","year":"2002","journal-title":"Nature Reviews Neuroscience"},{"issue":"1","key":"2024082018250840100_bib18","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1002\/cne.903000103","article-title":"Topography of ganglion cells in human retina","volume":"300","author":"Curcio","year":"1990","journal-title":"Journal of Comparative Neurology"},{"issue":"4","key":"2024082018250840100_bib19","doi-asserted-by":"publisher","first-page":"497","DOI":"10.1002\/cne.902920402","article-title":"Human photoreceptor topography","volume":"292","author":"Curcio","year":"1990","journal-title":"Journal of Comparative Neurology"},{"key":"2024082018250840100_bib20","author":"Dapello","year":"2020","journal-title":"Simulating a primary visual cortex at the front of CNNs improves robustness to image perturbations"},{"key":"2024082018250840100_bib21","doi-asserted-by":"crossref","first-page":"248","DOI":"10.1109\/CVPR.2009.5206848","article-title":"Imagenet: A large- scale hierarchical image database","volume-title":"Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition","author":"Deng","year":"2009"},{"issue":"1","key":"2024082018250840100_bib22","doi-asserted-by":"publisher","first-page":"219","DOI":"10.1113\/jphysiol.1984.sp015498","article-title":"Spatial and temporal contrast sensitivities of neurones in lateral geniculate nucleus of macaque","volume":"357","author":"Derrington","year":"1984","journal-title":"Journal of Physiology"},{"issue":"12","key":"2024082018250840100_bib23","doi-asserted-by":"publisher","first-page":"1827","DOI":"10.1016\/0042-6989(95)00294-4","article-title":"Saccade target selection and object recognition: Evidence for a common attentional mechanism","volume":"36","author":"Deubel","year":"1996","journal-title":"Vision Research"},{"key":"2024082018250840100_bib24","author":"Deza","year":"2020","journal-title":"Emergent properties of foveated perceptual systems."},{"issue":"3","key":"2024082018250840100_bib25","doi-asserted-by":"publisher","first-page":"415","DOI":"10.1016\/j.neuron.2012.01.010","article-title":"How does the brain solve visual object recognition?","volume":"73","author":"DiCarlo","year":"2012","journal-title":"Neuron"},{"issue":"6","key":"2024082018250840100_bib26","doi-asserted-by":"publisher","first-page":"621","DOI":"10.1089\/cpb.2004.7.621","article-title":"Gaze-contingent displays: A review","volume":"7","author":"Duchowski","year":"2004","journal-title":"Cyberpsychology and Behavior"},{"key":"2024082018250840100_bib27","author":"Elsayed","year":"2018","journal-title":"Adversarial examples that fool both computer vision and time-limited humans."},{"key":"2024082018250840100_bib28","author":"Esteves","year":"2017","journal-title":"Polar transformer networks"},{"key":"2024082018250840100_bib29","first-page":"4768","article-title":"Spatiotemporal multiplier networks for video action recognition","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Feichtenhofer","year":"2017"},{"issue":"9","key":"2024082018250840100_bib30","doi-asserted-by":"publisher","first-page":"1195","DOI":"10.1038\/nn.2889","article-title":"Metamers of the ventral stream","volume":"14","author":"Freeman","year":"2011","journal-title":"Nature Neuroscience"},{"issue":"4","key":"2024082018250840100_bib31","doi-asserted-by":"publisher","first-page":"519","DOI":"10.1002\/cne.902010405","article-title":"Visual topography of V2 in the macaque","volume":"201","author":"Gattass","year":"1981","journal-title":"Journal of Comparative Neurology"},{"issue":"6","key":"2024082018250840100_bib32","doi-asserted-by":"publisher","first-page":"1831","DOI":"10.1523\/JNEUROSCI.08-06-01831.1988","article-title":"Visuotopic organization and extent of V3 and V4 of the macaque","volume":"8","author":"Gattass","year":"1988","journal-title":"Journal of Neuroscience"},{"issue":"1","key":"2024082018250840100_bib33","doi-asserted-by":"publisher","first-page":"62","DOI":"10.1364\/JOSAA.3.000062","article-title":"Sampling-theory analysis of spatial vision","volume":"3","author":"Geisler","year":"1986","journal-title":"JOSA A"},{"key":"2024082018250840100_bib34","author":"Goodfellow","year":"2014","journal-title":"Explaining and harnessing adversarial examples."},{"key":"2024082018250840100_bib35","author":"Gu","year":"2014","journal-title":"Towards deep neural network architectures robust to adversarial examples."},{"key":"2024082018250840100_bib36","doi-asserted-by":"publisher","first-page":"67200","DOI":"10.1109\/ACCESS.2019.2918732","article-title":"Global-local attention network for aerial scene classification","volume":"7","author":"Guo","year":"2019","journal-title":"IEEE Access"},{"key":"2024082018250840100_bib37","article-title":"Finding biological plausibility for adversarially robust features via metameric tasks","author":"Harrington","year":"2021","journal-title":"Proceedings of the SVRHM 2021 Workshop@ NeurIPS"},{"key":"2024082018250840100_bib38","article-title":"Adversarial example defense: Ensembles of weak defenses are not strong","volume-title":"Proceedings of the 11th USENIX Workshop on Offensive Technologies","author":"He","year":"2017"},{"key":"2024082018250840100_bib39","first-page":"535","article-title":"Neural networks with recurrent generative feedback","volume-title":"Advances in neural information processing systems","author":"Huang","year":"2020"},{"issue":"3","key":"2024082018250840100_bib40","doi-asserted-by":"publisher","first-page":"194","DOI":"10.1038\/35058500","article-title":"Computational modelling of visual attention","volume":"2","author":"Itti","year":"2001","journal-title":"Nature Reviews Neuroscience"},{"issue":"11","key":"2024082018250840100_bib41","doi-asserted-by":"publisher","first-page":"1254","DOI":"10.1109\/34.730558","article-title":"A model of saliency-based visual attention for rapid scene analysis","volume":"20","author":"Itti","year":"1998","journal-title":"IEEE Transactions on Pattern Analysis and Machine Intelligence"},{"key":"2024082018250840100_bib42","author":"Jabbireddy","year":"2022","journal-title":"Foveated rendering: Motivation, taxonomy, and research directions"},{"key":"2024082018250840100_bib43","first-page":"4651","article-title":"Perceiver: General perception with iterative attention","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Jaegle","year":"2021"},{"key":"2024082018250840100_bib44","first-page":"1072","article-title":"Salicon: Saliency in context","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Jiang","year":"2015"},{"key":"2024082018250840100_bib45","author":"Jonnalagadda","year":"2021","journal-title":"Foveater: Foveated transformer for image classification."},{"issue":"6","key":"2024082018250840100_bib46","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3355089.3356557","article-title":"Deepfovea: Neural reconstruction for foveated rendering and video compression using learned statistics of natural videos","volume":"38","author":"Kaplanyan","year":"2019","journal-title":"ACM Transactions on Graphics"},{"issue":"6","key":"2024082018250840100_bib47","doi-asserted-by":"publisher","first-page":"974","DOI":"10.1038\/s41593-019-0392-5","article-title":"Evidence that recurrent circuits are critical to the ventral stream\u2019s execution of core object recognition behavior","volume":"22","author":"Kar","year":"2019","journal-title":"Nature Neuroscience"},{"issue":"43","key":"2024082018250840100_bib48","doi-asserted-by":"publisher","first-page":"21854","DOI":"10.1073\/pnas.1905544116","article-title":"Recurrence is required to capture the representational dynamics of the human visual system","volume":"116","author":"Kietzmann","year":"2019","journal-title":"Proceedings of the National Academy of Sciences"},{"key":"2024082018250840100_bib49","author":"Kingma","year":"2014","journal-title":"Adam: A method for stochastic optimization."},{"key":"2024082018250840100_bib50","doi-asserted-by":"crossref","first-page":"115","DOI":"10.1007\/978-94-009-3833-5_5","article-title":"Shifts in selective visual attention: Towards the underlying neural circuitry","volume-title":"Matters of intelligence","author":"Koch","year":"1987"},{"issue":"6","key":"2024082018250840100_bib51","doi-asserted-by":"publisher","first-page":"510","DOI":"10.1016\/j.cub.2012.01.050","article-title":"Temporal encoding of spatial information during active visual fixation","volume":"22","author":"Kuang","year":"2012","journal-title":"Current Biology"},{"issue":"11","key":"2024082018250840100_bib52","doi-asserted-by":"publisher","first-page":"571","DOI":"10.1016\/S0166-2236(00)01657-X","article-title":"The distinct modes of vision offered by feedforward and recurrent processing","volume":"23","author":"Lamme","year":"2000","journal-title":"Trends in Neurosciences"},{"issue":"4","key":"2024082018250840100_bib53","doi-asserted-by":"publisher","first-page":"10","DOI":"10.1002\/j.2326-1951.1976.tb01231.x","article-title":"On seeing sidelong","volume":"16","author":"Lettvin","year":"1976","journal-title":"Sciences"},{"key":"2024082018250840100_bib54","first-page":"9215","article-title":"Tell me where to look: Guided attention inference network","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Li","year":"2018"},{"key":"2024082018250840100_bib55","first-page":"740","article-title":"Microsoft COCO: Common objects in context","author":"Lin","year":"2014","journal-title":"European Conference on Computer Vision"},{"issue":"5","key":"2024082018250840100_bib56","doi-asserted-by":"publisher","first-page":"552","DOI":"10.1016\/S0960-9822(95)00108-4","article-title":"Shape representation in the inferior temporal cortex of monkeys","volume":"5","author":"Logothetis","year":"1995","journal-title":"Current Biology"},{"key":"2024082018250840100_bib57","author":"Luo","year":"2015","journal-title":"Foveation-based mechanisms alleviate adversarial examples"},{"key":"2024082018250840100_bib58","author":"Madry","year":"2017","journal-title":"Towards deep learning models resistant to adversarial attacks."},{"issue":"1","key":"2024082018250840100_bib59","doi-asserted-by":"publisher","first-page":"369","DOI":"10.1146\/annurev.ne.16.030193.002101","article-title":"How parallel are the primate visual pathways?","volume":"16","author":"Merigan","year":"1993","journal-title":"Annual Review of Neuroscience"},{"key":"2024082018250840100_bib60","author":"Min","year":"2022","journal-title":"Peripheral vision transformer."},{"key":"2024082018250840100_bib61","doi-asserted-by":"publisher","first-page":"414","DOI":"10.1016\/0166-2236(83)90190-X","article-title":"Object vision and spatial vision: Two cortical pathways","volume":"6","author":"Mishkin","year":"1983","journal-title":"Trends in Neurosciences"},{"key":"2024082018250840100_bib62","author":"Mnih","year":"2014","journal-title":"Recurrent models of visual attention."},{"issue":"7031","key":"2024082018250840100_bib63","doi-asserted-by":"publisher","first-page":"387","DOI":"10.1038\/nature03390","article-title":"Optimal eye movement strategies in visual search","volume":"434","author":"Najemnik","year":"2005","journal-title":"Nature"},{"key":"2024082018250840100_bib64","doi-asserted-by":"crossref","first-page":"582","DOI":"10.1109\/SP.2016.41","article-title":"Distillation as a defense to adversarial perturbations against deep neural networks","volume-title":"Proceedings of the 2016 IEEE Symposium on Security and Privacy","author":"Papernot","year":"2016"},{"key":"2024082018250840100_bib65","first-page":"57","article-title":"Gaze-contingent real-time simulation of arbitrary visual fields","volume":"7","author":"Perry","year":"2002","journal-title":"Human Vision and Electronic Imaging"},{"key":"2024082018250840100_bib66","first-page":"531","article-title":"Components of visual orienting","volume":"32","author":"Posner","year":"1984","journal-title":"Attention and Performance X: Control of Language Processes"},{"key":"2024082018250840100_bib67","doi-asserted-by":"publisher","first-page":"108083","DOI":"10.1016\/j.visres.2022.108083","article-title":"Human peripheral blur is optimal for object recognition","volume":"200","author":"Pramod","year":"2022","journal-title":"Vision Research"},{"issue":"11","key":"2024082018250840100_bib68","doi-asserted-by":"publisher","first-page":"1019","DOI":"10.1038\/14819","article-title":"Hierarchical models of object recognition in cortex","volume":"2","author":"Riesenhuber","year":"1999","journal-title":"Nature Neuroscience"},{"key":"2024082018250840100_bib69","doi-asserted-by":"publisher","first-page":"146","DOI":"10.1007\/s00221-003-1588-0","article-title":"Two different streams form the dorsal visual system: Anatomy and functions","volume":"153","author":"Rizzolatti","year":"2003","journal-title":"Experimental Brain Research"},{"key":"2024082018250840100_bib70","doi-asserted-by":"publisher","first-page":"437","DOI":"10.1146\/annurev-vision-082114-035733","article-title":"Capabilities and limitations of peripheral vision","volume":"2","author":"Rosenholtz","year":"2016","journal-title":"Annual Review of Vision Science"},{"key":"2024082018250840100_bib71","author":"Sermanet","year":"2014","journal-title":"Attention for fine-grained categorization."},{"key":"2024082018250840100_bib72","article-title":"Two-stream convolutional networks for action recognition in videos","volume-title":"Advances in neural information processing systems","author":"Simonyan","year":"2014"},{"issue":"12","key":"2024082018250840100_bib73","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1167\/jov.20.12.2","article-title":"A review of interactions between peripheral and foveal vision","volume":"20","author":"Stewart","year":"2020","journal-title":"Journal of Vision"},{"key":"2024082018250840100_bib74","author":"Szegedy","year":"2013","journal-title":"Intriguing properties of neural networks."},{"key":"2024082018250840100_bib75","first-page":"15539","article-title":"Fovea: Foveated image magnification for autonomous navigation","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision","author":"Thavamani","year":"2021"},{"issue":"8","key":"2024082018250840100_bib76","doi-asserted-by":"publisher","first-page":"1524","DOI":"10.1364\/JOSAA.4.001524","article-title":"Retinal limits to the detection and resolution of gratings","volume":"4","author":"Thibos","year":"1987","journal-title":"JOSA A"},{"issue":"6582","key":"2024082018250840100_bib77","doi-asserted-by":"publisher","first-page":"520","DOI":"10.1038\/381520a0","article-title":"Speed of processing in the human visual system","volume":"381","author":"Thorpe","year":"1996","journal-title":"Nature"},{"key":"2024082018250840100_bib78","first-page":"5025","article-title":"Adversarial risk and the dangers of evaluating against weak attacks","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Uesato","year":"2018"},{"key":"2024082018250840100_bib79","first-page":"2135","article-title":"Biologically inspired mechanisms for adversarial robustness","volume-title":"Advances in neural information processing systems","author":"Vuyyuru","year":"2020"},{"key":"2024082018250840100_bib80","author":"Wang","year":"2021","journal-title":"On the use of cortical magnification and saccades as biological proxies for data augmentation"},{"key":"2024082018250840100_bib81","first-page":"20","article-title":"Temporal segment networks: Towards good practices for deep action recognition","volume-title":"Proceedings of the European Conference on Computer Vision","author":"Wang","year":"2016"},{"key":"2024082018250840100_bib82","author":"Wang","year":"2020","journal-title":"Glance and focus: A dynamic approach to reducing spatial redundancy in image classification."},{"key":"2024082018250840100_bib83","doi-asserted-by":"publisher","first-page":"87","DOI":"10.1007\/s10339-016-0781-6","article-title":"Scanpath estimation based on foveated image saliency","volume":"18","author":"Wang","year":"2017","journal-title":"Cognitive Processing"},{"issue":"7","key":"2024082018250840100_bib84","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1167\/14.7.15","article-title":"A formula for human retinal ganglion cell receptive field density as a function of visual field location","volume":"14","author":"Watson","year":"2014","journal-title":"Journal of Vision"},{"issue":"1","key":"2024082018250840100_bib85","first-page":"75","article-title":"Implementations and implications of foveated vision","volume":"2","author":"Weber","year":"2009","journal-title":"Recent Patents on Computer Science"},{"key":"2024082018250840100_bib86","first-page":"1568","article-title":"Object region mining with adversarial erasing: A simple classification to semantic segmentation approach","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Wei","year":"2017"},{"key":"2024082018250840100_bib87","doi-asserted-by":"publisher","first-page":"472","DOI":"10.3389\/fpsyg.2012.00472","article-title":"Effects of peripheral visual field loss on eye movements during visual search","volume":"3","author":"Wiecek","year":"2012","journal-title":"Frontiers in Psychology"},{"issue":"3\u20134","key":"2024082018250840100_bib88","doi-asserted-by":"crossref","first-page":"229","DOI":"10.1007\/BF00992696","article-title":"Simple statistical gradient-following algorithms for connectionist reinforcement learning","volume":"8","author":"Williams","year":"1992","journal-title":"Machine Learning"},{"key":"2024082018250840100_bib89","author":"Xie","year":"2017","journal-title":"Mitigating adversarial effects through randomization."},{"key":"2024082018250840100_bib90","first-page":"2048","article-title":"Show, attend and tell: Neural image caption generation with visual attention","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Xu","year":"2015"},{"key":"2024082018250840100_bib91","first-page":"2921","article-title":"Learning deep features for discriminative localization","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Zhou","year":"2016"},{"key":"2024082018250840100_bib92","first-page":"9483","article-title":"Towards robust image classification using sequential attention models","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Zoran","year":"2020"}],"container-title":["Neural Computation"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/neco\/article-pdf\/36\/9\/1713\/2465936\/neco_a_01688.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/neco\/article-pdf\/36\/9\/1713\/2465936\/neco_a_01688.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,8,20]],"date-time":"2024-08-20T18:27:11Z","timestamp":1724178431000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/neco\/article\/36\/9\/1713\/123687\/Human-Eyes-Inspired-Recurrent-Neural-Networks-Are"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8,19]]},"references-count":92,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2024,8,19]]},"published-print":{"date-parts":[[2024,8,19]]}},"URL":"https:\/\/doi.org\/10.1162\/neco_a_01688","relation":{},"ISSN":["0899-7667","1530-888X"],"issn-type":[{"value":"0899-7667","type":"print"},{"value":"1530-888X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2024,9]]},"published":{"date-parts":[[2024,8,19]]}}}