Deep Residual Network Predicts Cortical Representation and Organization of Visual Features for Rapid Categorization

Wen, Haiguang; Shi, Junxing; Chen, Wei; Liu, Zhongming

doi:10.1038/s41598-018-22160-9

Cited by 67 publications

(75 citation statements)

References 79 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent research in machine learning has shown that deep convolutional neural networks (DCNNs) [2] can perform invariant object categorization with almost human-level accuracy [3], and that their network representations are similar to the brain's [4][5] [6]. DCNNs are therefore very important as models of visual hierarchy [7][8] [9], though understanding their operational capabilities and design principles remain a significant challenge [10].…”

Section: Introductionmentioning

confidence: 99%

Separability and Geometry of Object Manifolds in Deep Neural Networks

Cohen

Chung

Lee³

et al. 2019

Preprint

View full text Add to dashboard Cite

Stimuli are represented in the brain by the collective population responses of sensory neurons, and an object presented under varying conditions gives rise to a collection of neural population responses called an "object manifold." Changes in the object representation along a hierarchical sensory system are associated with changes in the geometry of those manifolds, and recent theoretical progress connects this geometry with "classification capacity," a quantitative measure of the ability to support object classification. Deep neural networks trained on object classification tasks are a natural testbed for the applicability of this relation. We show how classification capacity improves along the hierarchies of deep neural networks with different architectures. We demonstrate that changes in the geometry of the associated object manifolds underlie this improved capacity, and shed light on the functional roles different levels in the hierarchy play to achieve it, through orchestrated reduction of manifolds' radius, dimensionality and inter-manifold correlations.

show abstract

Section: Introductionmentioning

confidence: 99%

Separability and Geometry of Object Manifolds in Deep Neural Networks

Cohen

Chung

Lee³

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…An alternative approach to using multi-voxel patterns is to model feature-selectivity at the individual voxel level (Naselaris et al, 2011). While this approach might be more sensitive to more fine-grained selectivity, it is striking that studies using this approach have primarily revealed smooth gradients across visual cortex that largely seem to reflect the large-scale category-selective organization (Huth et al, 2012; Wen et al, 2018) with evidence for a limited number of functional sub-domains (Çukur et al, 2013, 2016).…”

Section: Discussionmentioning

confidence: 99%

Similarity judgments and cortical visual responses reflect different properties of object and scene categories in naturalistic images

King

Groen

Steel³

et al. 2018

Preprint

View full text Add to dashboard Cite

1 2 Numerous factors have been reported to underlie the representation of complex images in high-3 level human visual cortex, including categories (e.g. faces, objects, scenes), animacy, and real-4 world size, but the extent to which this organization is reflected in behavioral judgments of real-5 world stimuli is unclear. Here, we compared representations derived from explicit similarity 6 judgments and ultra-high field (7T) fMRI of human visual cortex for multiple exemplars of a diverse 7 set of naturalistic images from 48 object and scene categories. Behavioral judgements revealed a 8 coarse division between man-made (including humans) and natural (including animals) images, 9 with clear groupings of conceptually-related categories (e.g. transportation, animals), while these 10 conceptual groupings were largely absent in the fMRI representations. Instead, fMRI responses 11 tended to reflect a separation of both human and non-human faces/bodies from all other categories. 12 This pattern yielded a statistically significant, but surprisingly limited correlation between the two 13 representational spaces. Further, comparison of the behavioral and fMRI representational spaces 14 with those derived from the layers of a deep neural network (DNN) showed a strong 15 correspondence with behavior in the top-most layer and with fMRI in the mid-level layers. These 16 results suggest that there is no simple mapping between responses in high-level visual cortex and 17 behavior -each domain reflects different visual properties of the images and responses in high-18 level visual cortex may correspond to intermediate stages of processing between basic visual 19 features and the conceptual categories that dominate the behavioral response. 20 29 highlight that great care must be taken in mapping the response of visual cortex onto behavior, 30 which clearly reflect different information. 31 50 51 Determining how responses in high-level visual cortex relate to behavior is critical for elucidating 52 the functional significance of these regions. For tasks such as identification and categorization, 53 relevant information has been reported in the responses of lOTC and vTC (Kravitz et al., 2013; 54

show abstract

“…There are many different architectures of CNNs in the literature,with associated advantages and drawbacks. In the current study, we used the deep residual network (ResNet) [25], a model which has shown great performance in image classification problems including medical image analysis [8], [26]. Although it has been shown that CNNs with more convolutional layers achieve the most accurate results, simply stacking more convolutional layers will not lead to better performance.…”

Section: A Model Developmentmentioning

confidence: 99%

CeliacNet: Celiac Disease Severity Diagnosis on Duodenal Histopathological Images Using Deep Residual Networks

Sali

Ehsan

Kowsari

et al. 2019

2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

View full text Add to dashboard Cite

Celiac Disease (CD) is a chronic autoimmune disease that affects the small intestine in genetically predisposed children and adults. Gluten exposure triggers an inflammatory cascade which leads to compromised intestinal barrier function. If this enteropathy is unrecognized, this can lead to anemia, decreased bone density, and, in longstanding cases, intestinal cancer. The prevalence of the disorder is 1% in the United States. An intestinal (duodenal) biopsy is considered the "gold standard" for diagnosis. The mild CD might go unnoticed due to non-specific clinical symptoms or mild histologic features. In our current work, we trained a model based on deep residual networks to diagnose CD severity using a histological scoring system called the modified Marsh score. The proposed model was evaluated using an independent set of 120 whole slide images from 15 CD patients and achieved an AUC greater than 0.96 in all classes. These results demonstrate the diagnostic power of the proposed model for CD severity classification using histological images.

show abstract

Deep Residual Network Predicts Cortical Representation and Organization of Visual Features for Rapid Categorization

Cited by 67 publications

References 79 publications

Separability and Geometry of Object Manifolds in Deep Neural Networks

Separability and Geometry of Object Manifolds in Deep Neural Networks

Similarity judgments and cortical visual responses reflect different properties of object and scene categories in naturalistic images

CeliacNet: Celiac Disease Severity Diagnosis on Duodenal Histopathological Images Using Deep Residual Networks

Contact Info

Product

Resources

About