scholarly journals Visual features as stepping stones toward semantics: Explaining object similarity in IT and perception with non-negative least squares.

2015 ◽  
Author(s):  
Kamila M Jozwik ◽  
Nikolaus Kriegeskorte ◽  
Marieke Mur

Object similarity, in brain representations and conscious perception, must reflect a combination of the visual appearance of the objects on the one hand and the categories the objects belong to on the other. Indeed, visual object features and category membership have each been shown to contribute to the object representation in human inferior temporal (IT) cortex, as well as to object-similarity judgments. However, the explanatory power of features and categories has not been directly compared. Here, we investigate whether the IT object representation and similarity judgments are best explained by a categorical or a feature-based model. We use rich models (> 100 dimensions) generated by human observers for a set of 96 real-world object images. The categorical model consists of a hierarchically nested set of category labels (such as 'human', 'mammal', 'animal'). The feature model includes both object parts (such as 'eye', 'tail', 'handle') and other descriptive features (such as 'circular', 'green', 'stubbly'). We used non-negative least squares to fit the models to the brain representations (estimated from functional magnetic resonance imaging data) and to similarity judgments. Model performance was estimated on held-out images not used in fitting. Both models explained significant variance in IT and the amounts explained were not significantly different. The combined model did not explain significant additional IT variance, suggesting that it is the shared model variance (features correlated with categories, categories correlated with features) that best explains IT. The similarity judgments were almost fully explained by the categorical model, which explained significantly more variance than the feature-based model. The combined model did not explain significant additional variance in the similarity judgments. Our findings suggest that IT uses features that help to distinguish categories as stepping stones toward a semantic representation. Similarity judgments contain additional categorical variance that is not explained by visual features, reflecting a higher-level more purely semantic representation. Keywords: object vision, categories, features, human inferior temporal cortex, fMRI, representational similarity analysis

2013 ◽  
Vol 4 ◽  
Author(s):  
Marieke Mur ◽  
Mirjam Meys ◽  
Jerzy Bodurka ◽  
Rainer Goebel ◽  
Peter A. Bandettini ◽  
...  

Psihologija ◽  
2010 ◽  
Vol 43 (2) ◽  
pp. 155-165 ◽  
Author(s):  
Vanja Kovic ◽  
Kim Plunkett ◽  
Gert Westermann

In this paper we present an ERP study examining the underlying nature of semantic representation of animate and inanimate objects. Time-locking ERP signatures to the onset of auditory stimuli we found topological similarities in animate and inanimate object processing. Moreover, we found no difference between animates and inanimates in the N400 amplitude, when mapping more specific to more general representation (visual to auditory stimuli). These studies provide further evidence for the theory of unitary semantic organization, but no support for the feature-based prediction of segregated conceptual organization. Further comparisons of animate vs. inanimate matches and within-vs. between-category mismatches revealed following results: processing of animate matches elicited more positivity than processing of inanimates within the N400 time-window; also, inanimate mismatches elicited a stronger N400 than did animate mismatches. Based on these findings we argue that one of the possible explanations for finding different and sometimes contradictory results in the literature regarding processing and representations of animates and inanimates in the brain could lie in the variability of selected items within each of the categories, that is, homogeneity of the categories.


2018 ◽  
Vol 24 (6) ◽  
pp. 861-886 ◽  
Author(s):  
ABDULGABBAR SAIF ◽  
UMMI ZAKIAH ZAINODIN ◽  
NAZLIA OMAR ◽  
ABDULLAH SAEED GHAREB

AbstractSemantic measures are used in handling different issues in several research areas, such as artificial intelligence, natural language processing, knowledge engineering, bioinformatics, and information retrieval. Hierarchical feature-based semantic measures have been proposed to estimate the semantic similarity between two concepts/words depending on the features extracted from a semantic taxonomy (hierarchy) of a given lexical source. The central issue in these measures is the constant weighting assumption that all elements in the semantic representation of the concept possess the same relevance. In this paper, a new weighting-based semantic similarity measure is proposed to address the issues in hierarchical feature-based measures. Four mechanisms are introduced to weigh the degree of relevance of features in the semantic representation of a concept by using topological parameters (edge, depth, descendants, and density) in a semantic taxonomy. With the semantic taxonomy of WordNet, the proposed semantic measure is evaluated for word semantic similarity in four gold-standard datasets. Experimental results show that the proposed measure outperforms hierarchical feature-based semantic measures in all the datasets. Comparison results also imply that the proposed measure is more effective than information-content measures in measuring semantic similarity.


Author(s):  
Chong Wang ◽  
Zheng-Jun Zha ◽  
Dong Liu ◽  
Hongtao Xie

High-level semantic knowledge in addition to low-level visual cues is essentially crucial for co-saliency detection. This paper proposes a novel end-to-end deep learning approach for robust co-saliency detection by simultaneously learning highlevel group-wise semantic representation as well as deep visual features of a given image group. The inter-image interaction at semantic-level as well as the complementarity between group semantics and visual features are exploited to boost the inferring of co-salient regions. Specifically, the proposed approach consists of a co-category learning branch and a co-saliency detection branch. While the former is proposed to learn group-wise semantic vector using co-category association of an image group as supervision, the latter is to infer precise co-salient maps based on the ensemble of group semantic knowledge and deep visual cues. The group semantic vector is broadcasted to each spatial location of multi-scale visual feature maps and is used as a top-down semantic guidance for boosting the bottom-up inferring of co-saliency. The co-category learning and co-saliency detection branches are jointly optimized in a multi-task learning manner, further improving the robustness of the approach. Moreover, we construct a new large-scale co-saliency dataset COCO-SEG to facilitate research of co-saliency detection. Extensive experimental results on COCO-SEG and a widely used benchmark Cosal2015 have demonstrated the superiority of the proposed approach as compared to the state-of-the-art methods.


2018 ◽  
Author(s):  
Angus F. Chapman ◽  
Viola S. Störmer

Theories of visual attention differ in what they define as the core unit of selection. Feature-based theories emphasize the importance of visual features (e.g., color, size, motion), demonstrated through enhancement of attended features across the visual field, while object-based theories propose that attention enhances all features belonging to the same object. Here we test how within-object enhancement of features interacts with spatially global effects of feature-based attention. Participants attended a set of colored dots (moving coherently upwards or downwards) to detect brief luminance decreases, while simultaneously detecting speed changes in another set of dots in the opposite visual field. Participants had higher speed detection rates for the dot array that matched the motion direction of the attended color array, although motion direction was entirely task-irrelevant. This effect persisted even when it was detrimental for task performance. Overall, these results indicate that task-irrelevant object features are enhanced globally, surpassing object boundaries.


2011 ◽  
Vol 105 (3) ◽  
pp. 1258-1265 ◽  
Author(s):  
Vivian M. Ciaramitaro ◽  
Jude F. Mitchell ◽  
Gene R. Stoner ◽  
John H. Reynolds ◽  
Geoffrey M. Boynton

Faced with an overwhelming amount of sensory information, we are able to prioritize the processing of select spatial locations and visual features. The neuronal mechanisms underlying such spatial and feature-based selection have been studied in considerable detail. More recent work shows that attention can also be allocated to objects, even spatially superimposed objects composed of dynamically changing features that must be integrated to create a coherent object representation. Much less is known about the mechanisms underlying such object-based selection. Our goal was to investigate behavioral and neuronal responses when attention was directed to one of two objects, specifically one of two superimposed transparent surfaces, in a task designed to preclude space-based and feature-based selection. We used functional magnetic resonance imaging (fMRI) to measure changes in blood oxygen level-dependent (BOLD) signals when attention was deployed to one or the other surface. We found that visual areas V1, V2, V3, V3A, and MT+ showed enhanced BOLD responses to translations of an attended relative to an unattended surface. These results reveal that visual areas as early as V1 can be modulated by attending to objects, even objects defined by dynamically changing elements. This provides definitive evidence in humans that early visual areas are involved in a seemingly high-order process. Furthermore, our results suggest that these early visual areas may participate in object-specific feature “binding,” a process that seemingly must occur for an object or a surface to be the unit of attentional selection.


2007 ◽  
Vol 28 ◽  
pp. 349-391 ◽  
Author(s):  
S. R. Jodogne ◽  
J. H. Piater

In this paper we present a general, flexible framework for learning mappings from images to actions by interacting with the environment. The basic idea is to introduce a feature-based image classifier in front of a reinforcement learning algorithm. The classifier partitions the visual space according to the presence or absence of few highly informative local descriptors that are incrementally selected in a sequence of attempts to remove perceptual aliasing. We also address the problem of fighting overfitting in such a greedy algorithm. Finally, we show how high-level visual features can be generated when the power of local descriptors is insufficient for completely disambiguating the aliased states. This is done by building a hierarchy of composite features that consist of recursive spatial combinations of visual features. We demonstrate the efficacy of our algorithms by solving three visual navigation tasks and a visual version of the classical ``Car on the Hill'' control problem.


Sign in / Sign up

Export Citation Format

Share Document