Figure - available from: Computational Intelligence and Neuroscience
This content is subject to copyright. Terms and conditions apply.
Taxonomic structure of recognition used in this paper. A i and P i refer to the subcategories of animal and plant correspondingly.
Source publication
It has been argued that concepts can be perceived at three main levels of abstraction. Generally, in a recognition system, object categories can be viewed at three levels of taxonomic hierarchy which are known as superordinate, basic, and subordinate levels. For instance, "horse" is a member of subordinate level which belongs to basic level of "ani...
Similar publications
Recent studies provided an increasingly detailed understanding of how visual objects categories like faces or bodies are represented in the brain. What is less clear is how a given task impacts the representation of the object category and of its attributes. Using (fMRI) we measured BOLD responses while participants viewed whole body expressions an...
Citations
... Based on object features, the idea of hierarchical classification classifies the objects into different clusters, and each cluster has high similarities. As the number of levels increases, the similarity in each cluster increases [21]. Goel et al. [22] create a binary tree hierarchy using a max-margin clustering strategy. ...
In computer vision, fine-grained classification has become an important issue in recognizing objects with slight visual differences. Usually, it is challenging to generate good performance when solving fine-grained classification problems using traditional convolutional neural networks. To improve the accuracy and training time of convolutional neural networks in solving fine-grained classification problems, this paper proposes a tree-structured framework by eliminating the effect of differences between clusters. The contributions of the proposed method include the following three aspects: (1) a self-supervised method that automatically creates a classification tree, eliminating the need for manual labeling; (2) a machine-learning matcher which determines the cluster to which an item belongs, minimizing the impact of inter-cluster variations on classification; and (3) a pruning criterion which filters the tree-structured classifier, retaining only the models with superior classification performance. The experimental evaluation of the proposed tree-structured framework demonstrates its effectiveness in reducing training time and improving the accuracy of fine-grained classification across various datasets in comparison with conventional convolutional neural network models. Specifically, for the CUB 200 2011, FGVC aircraft, and Stanford car datasets, the proposed method achieves a reduction in training time of 32.91%, 35.87%, and 14.48%, and improves the accuracy of fine-grained classification by 1.17%, 2.01%, and 0.59%, respectively.
... The taxonomic tree was derived from the taxonomic categorization (Sadeghi et al., 2015) categorizing objects and visual features to perform prominent roles in each category. The tree we created is based on the phylogenetic tree of CUB-200-2011 birds' information. ...
... Based on what was mentioned, our proposed method tries to build phylogenetic tree birds and put each image in the correct position on the tree. The idea of proposing a method is based on the taxonomic categorization (Sadeghi et al., 2015), used for categorizing objects where visual features perform a prominent role. It presents a hierarchy with multiple classification levels. ...
Fine-grained categorization is one of the most challenging problems in machine vision. Recently, the presented methods have been based on convolutional neural networks, increasing the accuracy of classification very significantly. Inspired by these methods, we offer a new framework for fine-grained categorization. Our tree method, named “TMTCPT”, is based on the taxonomic categorization, phylogenetic tree, and convolutional neural network classifiers. The word “taxonomic” has been derived from “taxonomical categorization” that categorizes objects and visual features and performs a prominent role in this category. It presents a hierarchical categorization that leads to multiple classification levels; the first level includes the general visual features having the lowest similarity level, whereas the other levels include visual features strikingly similar, as they follow top-bottom hierarchy. The phylogenetic tree presents the phylogenetic information of organisms. The convolutional neural network classifiers can classify the categories precisely. In this study, the researchers created a tree to increase classification accuracy and evaluated the effectiveness of the method by examining it on the challenging CUB-200-2011 dataset. The study results demonstrated that the proposed method was efficient and robust. The average classification accuracy of the proposed method was 88.34%, being higher than those of all the previous methods.
... This impairment has highly reported in living object category rather than non-living objects (Warrington & Shallice, 1984). The most likely explanation for this evidence is the higher amount of visual correlation between semantically related items that can be found in living group than the non-living group (Sadeghi et al., 2015;Sadeghi, Ahmadabadi, & Araabi, 2013;Gonnerman et al., 1997). Based on Evolutionary reasons of survival, recognition of animal and plant object types is necessary for human-being (Caramazza & Shelton, 1998). ...
How humans can distinguish between general categories of objects? Are the subcategories of living things visually distinctive? In a number of semantic-category deficits, patients are good at making broad categorization but are unable to remember fine and specific details. It has been well accepted that general information about concepts are more robust to damages related to semantic memory. Results from patients with semantic memory disorders demonstrate the loss of ability in subcategory recognition. While bottom-up feature construction has been studied in detail, little attention has been served to top-down approach and the type of features that could account for general categorization. In this paper, we show that broad categories of animal and plant are visually distinguishable without processing textural information. To this aim, we utilize shape descriptors with an additional phase of feature learning. The results are evaluated with both supervised and unsupervised learning mechanisms. The obtained results demonstrate that global encoding of visual appearance of objects accounts for high discrimination between animal and plant object categories.
... Concepts can be perceived at superordinate, basic and subordinate levels of abstraction (e.g. Sadeghi et al., 2015) and they have also been subdivided in cognition studies into 'mass' superordinates and 'count' superordinates (Wisniewski et al., 1996). However the first group of multidisciplinary concepts, here termed superordinate can, for example, include the over-arching concepts recognised by the RGS as discussed earlier (place, space, scale, interdependence, physical and human processes, environmental interaction and sustainable development). ...
The word concept is widely used in physical geography but seldom defined. Developing from an earlier proposal of concept types in geomorphology, this paper considers a structure for categorising concepts in physical geography in the light of sciences and philosophy more generally. It reviews where our concepts derive from, and their relation to kinds, universals and categories, whilst also indicating the lack of an agreed clear distinction between them. Because an unstructured diversity of concepts has previously been proposed in physical geography by different authors, a new provisional hierarchy is constructed. This is cognisant of specific developments in a range of disciplines, including formal concept analysis, lattice theory and hierarchy theory. A ‘concept of concepts’ hierarchy of six categories is proposed in which multidisciplinary (superordinate, contextual meta-concepts) and fundamental, operational and ancillary categories provide a 6 × 5 framework. This enables attention to be focused on stimulating conceptual underpinnings that can be tested at different levels in future learning, teaching and research. This can support the formation of knowledge structures and monitoring procedures that keep in step with ways that are characterising other disciplines.
... However, from a theoretical perspective even the study of simple character recognition might provide important insights. Indeed, written symbols often have a complex spatial structure, [16], where lower-levels visual features such as edges and grating (e.g., [17] [18]) are successively combined in order to produce more abstract and invariant features, such as letter shapes [19]. On the one hand, it might be reasonable to assume that in bilingual individuals abstract symbols belonging to different languages are processed and represented in distinct regions of the visual cortex, which might facilitate word recognition since the constituting characters usually belong to a single script. ...
... Holistic object perception and representation is known as the most recognizable theory of mental representation which is consistent with Gestalt principal and has been applied in many object recognition tasks. In [30] shape based descriptors are utilized for classification of animal vs plant categories. The second group or part-concept group is marked with usage of specific components which can be served as diagnostic cues for identifying members of a category. ...
In this paper, I investigate conceptual categories derived from developmental processing in a deep neural network. The similarity matrices of deep representation at each layer of neural network are computed and compared with their raw representation. While the clusters generated by raw representation stand at the basic level of abstraction, conceptual categories obtained from deep representation shows a bottom-up transition procedure. Results demonstrate a developmental course of learning from specific to general level of abstraction through learned layers of representations in a deep belief network.
