scispace - formally typeset
Search or ask a question

Showing papers on "Representation (systemics) published in 2014"


Journal ArticleDOI
TL;DR: The proposed FDDL model is extensively evaluated on various image datasets, and it shows superior performance to many state-of-the-art dictionary learning methods in a variety of classification tasks.
Abstract: The employed dictionary plays an important role in sparse representation or sparse coding based image reconstruction and classification, while learning dictionaries from the training data has led to state-of-the-art results in image classification tasks. However, many dictionary learning models exploit only the discriminative information in either the representation coefficients or the representation residual, which limits their performance. In this paper we present a novel dictionary learning method based on the Fisher discrimination criterion. A structured dictionary, whose atoms have correspondences to the subject class labels, is learned, with which not only the representation residual can be used to distinguish different classes, but also the representation coefficients have small within-class scatter and big between-class scatter. The classification scheme associated with the proposed Fisher discrimination dictionary learning (FDDL) model is consequently presented by exploiting the discriminative information in both the representation residual and the representation coefficients. The proposed FDDL model is extensively evaluated on various image datasets, and it shows superior performance to many state-of-the-art dictionary learning methods in a variety of classification tasks.

474 citations


Book ChapterDOI
06 Sep 2014
TL;DR: This work learns a knowledge base (KB) using a Markov Logic Network (MLN) and shows that a diverse set of visual inference tasks can be done in this unified framework without training separate classifiers, including zero-shot affordance prediction and object recognition given human poses.
Abstract: Reasoning about objects and their affordances is a fundamental problem for visual intelligence. Most of the previous work casts this problem as a classification task where separate classifiers are trained to label objects, recognize attributes, or assign affordances. In this work, we consider the problem of object affordance reasoning using a knowledge base representation. Diverse information of objects are first harvested from images and other meta-data sources. We then learn a knowledge base (KB) using a Markov Logic Network (MLN). Given the learned KB, we show that a diverse set of visual inference tasks can be done in this unified framework without training separate classifiers, including zero-shot affordance prediction and object recognition given human poses.

252 citations



Book ChapterDOI
01 Nov 2014
TL;DR: This paper presents a novel visual representation, called orderlets, for real-time human action recognition with depth sensors, that is insensitive to small noise since an orderlet only depends on the comparative relationship among individual features.
Abstract: This paper presents a novel visual representation, called orderlets, for real-time human action recognition with depth sensors. An orderlet is a middle level feature that captures the ordinal pattern among a group of low level features. For skeletons, an orderlet captures specific spatial relationship among a group of joints. For a depth map, an orderlet characterizes a comparative relationship of the shape information among a group of subregions. The orderlet representation has two nice properties. First, it is insensitive to small noise since an orderlet only depends on the comparative relationship among individual features. Second, it is a frame-level representation thus suitable for real-time online action recognition. Experimental results demonstrate its superior performance on online action recognition and cross-environment action recognition.

172 citations


Journal ArticleDOI
TL;DR: The impact of the ontic vs. epistemic sets distinction in statistics is examined to show its importance because there is a risk of misusing basic notions and tools, such as conditioning, distance between sets, variance, regression, etc. when data are set-valued.

165 citations


Posted Content
TL;DR: This work proposes a model for fine-grained categorization that overcomes limitations by leveraging deep convolutional features computed on bottom-up region proposals, and learns whole-object and part detectors, enforces learned geometric constraints between them, and predicts a fine- grained category from a pose-normalized representation.
Abstract: Semantic part localization can facilitate fine-grained categorization by explicitly isolating subtle appearance differences associated with specific object parts. Methods for pose-normalized representations have been proposed, but generally presume bounding box annotations at test time due to the difficulty of object detection. We propose a model for fine-grained categorization that overcomes these limitations by leveraging deep convolutional features computed on bottom-up region proposals. Our method learns whole-object and part detectors, enforces learned geometric constraints between them, and predicts a fine-grained category from a pose-normalized representation. Experiments on the Caltech-UCSD bird dataset confirm that our method outperforms state-of-the-art fine-grained categorization methods in an end-to-end evaluation without requiring a bounding box at test time.

162 citations


Journal ArticleDOI
TL;DR: In this paper, based on a more effective representation of uncertainty, called D numbers, a new method is proposed for the EIA problem, and the assessment results of environmental impacts are expressed and modeled by D numbers.
Abstract: Environmental impact assessment (EIA) is a complex problem influenced by many aspects, such as environmental, social, economic, etc. Due to the involvement of human judgment, various uncertainties are introduced in the EIA process. One critical issue of EIA is the representation and handling of uncertain information. Many different theories are available to deal with uncertainty, however, deficiencies exist in these theories. In this paper, based on a more effective representation of uncertainty, called D numbers, a new method is proposed for the EIA problem. In the proposed method, the assessment results of environmental impacts are expressed and modeled by D numbers. An illustrative case is provided to demonstrate the effectiveness of the proposed method.

152 citations


Journal ArticleDOI
TL;DR: This paper analyzes the novel concept of object bank, a high-level image representation encoding object appearance and spatial location information in images, and demonstrates that object bank is a high level representation, from which it can easily discover semantic information of unknown images.
Abstract: It is a remarkable fact that images are related to objects constituting them. In this paper, we propose to represent images by using objects appearing in them. We introduce the novel concept of object bank (OB), a high-level image representation encoding object appearance and spatial location information in images. OB represents an image based on its response to a large number of pre-trained object detectors, or `object filters', blind to the testing dataset and visual recognition task. Our OB representation demonstrates promising potential in high level image recognition tasks. It significantly outperforms traditional low level image representations in image classification on various benchmark image datasets by using simple, off-the-shelf classification algorithms such as linear SVM and logistic regression. In this paper, we analyze OB in detail, explaining our design choice of OB for achieving its best potential on different types of datasets. We demonstrate that object bank is a high level representation, from which we can easily discover semantic information of unknown images. We provide guidelines for effectively applying OB to high level image recognition tasks where it could be easily compressed for efficient computation in practice and is very robust to various classifiers.

149 citations


Journal ArticleDOI
TL;DR: This paper highlights several misinterpretations that arise in the field of fractional systems analysis using a representation known in the literature as “state space description” and proposed alternative descriptions are proposed.
Abstract: This paper highlights several misinterpretations that arise in the field of fractional systems analysis using a representation known in the literature as "state space description" Given these misinterpretations, some results already published and based on this description are questionable Thus alternative descriptions are proposed

137 citations


Journal ArticleDOI
TL;DR: The solution is to recognise that the contents of intentions can be partially determined byThe contents of motor representations, which enables better understanding how intentions relate to actions.
Abstract: Are there distinct roles for intention and motor representation in explaining the purposiveness of action? Standard accounts of action assign a role to intention but are silent on motor representation. The temptation is to suppose that nothing need be said here because motor representation is either only an enabling condition for purposive action or else merely a variety of intention. This paper provides reasons for resisting that temptation. Some motor representations, like intentions, coordinate actions in virtue of representing outcomes; but, unlike intentions, motor representations cannot feature as premises or conclusions in practical reasoning. This implies that motor representation has a distinctive role in explaining the purposiveness of action. It also gives rise to a problem: were the roles of intention and motor representation entirely independent, this would impair effective action. It is therefore necessary to explain how intentions interlock with motor representations. The solution, we argue, is to recognise that the contents of intentions can be partially determined by the contents of motor representations. Understanding this content-determining relation enables better understanding how intentions relate to actions.

136 citations


02 Jul 2014
TL;DR: This work measured population responses in cat primary visual cortex using electrode arrays and found that normalization has profound effects on V1 population responses and is likely to shape the interpretation of these responses by higher cortical areas.

Book
17 Oct 2014
TL;DR: In this paper, the principles and practices of arts-related inquiry are outlined and case study examples are provided to address the concerns academics increasingly appear to be voicing about developing the scholarship and practice of arts related research.
Abstract: This book outlines the principles and practices of arts-related inquiry and provides both suggestions about conducting research in the field as well as case study examples. The ideas presented here have emerged from the authors’ own experiences of undertaking arts-related research and the challenges of implementing these approaches. The book therefore draws on personal research, practice and experience to address the concerns academics increasingly appear to be voicing about developing the scholarship and practice of arts-related research. There is a need for greater attention to, and clarity on, issues of theoretical positioning, methodology and methods when conducting robust and reputable arts-related research, which this book provides.

Proceedings Article
27 Jul 2014
TL;DR: It is shown that symbols that can represent the preconditions and effects of an agent's actions are both necessary and sufficient for high-level planning, which eliminates the symbol design problem when a representation must be constructed in advance and in principle enables an agent to autonomously learn its own symbolic representations.
Abstract: We consider the problem of constructing a symbolic description of a continuous, low-level environment for use in planning. We show that symbols that can represent the preconditions and effects of an agent's actions are both necessary and sufficient for high-level planning. This eliminates the symbol design problem when a representation must be constructed in advance, and in principle enables an agent to autonomously learn its own symbolic representations. The resulting representation can be converted into PDDL, a canonical high-level planning representation that enables very fast planning.

Book ChapterDOI
14 Jan 2014
TL;DR: The authors argue that the process of mentally simulating events so as to predict their outcome, a facility possessed by most people for common contexts, is extended and refined in a skilled scientist to become a sharp and crucial intuition that can be used in solving difficult, complex or extraordinary problems.
Abstract: Inasmuch as people are good at predicting the outcome of physical interactions in the world around them, why are they so bad at physics, even the branch of physics (mechanics) that deals with the interaction of everyday objects? I argue here that the process of mentally simulating events so as to predict their outcome, a facility possessed by most people for common contexts, is extended and refined in a skilled scientist to become a sharp and crucial intuition that can be used in solving difficult, complex or extraordinary problems. Novices, lacking this ex­ tended intuition, find such problems difficult.


Posted Content
22 Jun 2014
TL;DR: This work proposes to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network, and naturally supports object recognition from 2.5D depth map and also view planning for object recognition.
Abstract: 3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representations automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet -- a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks.

Proceedings ArticleDOI
01 Sep 2014
TL;DR: This work designs actions involving use of tools such as forks and knives that obtain haptic data containing information about the physical properties of the object, and presents a method to compactly represent the robot's beliefs about the object's properties using a generative model.
Abstract: Manipulation of complex deformable semi-solids such as food objects is an important skill for personal robots to have. In this work, our goal is to model and learn the physical properties of such objects. We design actions involving use of tools such as forks and knives that obtain haptic data containing information about the physical properties of the object. We then design appropriate features and use supervised learning to map these features to certain physical properties (hardness, plasticity, elasticity, tensile strength, brittleness, adhesiveness). Additionally, we present a method to compactly represent the robot's beliefs about the object's properties using a generative model, which we use to plan appropriate manipulation actions. We extensively evaluate our approach on a dataset including haptic data from 12 categories of food (including categories not seen before by the robot) obtained in 941 experiments. Our robot prepared a salad during 60 sequential robotic experiments where it made a mistake in only 4 instances.

Patent
14 Apr 2014
TL;DR: In this article, a virtual assistant ecosystem is presented, where one can instantiate or construct a customized virtual assistant when needed by capturing a digital representation of one or more objects, which can then be analyzed to determine the nature or type of the objects present.
Abstract: A virtual assistant ecosystem is presented. One can instantiate or construct a customized virtual assistant when needed by capturing a digital representation of one or more objects. A virtual assistant engine analyzes the digital representation to determine the nature or type of the objects present. The engine further obtains attributes for a desirable assistant based on the type of objects. Once the attributes are compiled the engine can then create the specific type of assistant required by the circumstances.

Patent
13 Mar 2014
TL;DR: In this article, various types of user feedback may be provided to a human user operator of the mobile device, and particular images may be selected for further analysis in various manners, such as generating and manipulating a computer model or other representation of the object from selected images.
Abstract: Techniques are described for analyzing images acquired via mobile devices in various ways, including to estimate measurements for one or more attributes of one or more objects in the images. For example, the described techniques may be used to measure the volume of a stockpile of material or other large object, based on images acquired via a mobile device that is carried by a human user as he or she passes around some or all of the object. During the acquisition of a series of digital images of an object of interest, various types of user feedback may be provided to a human user operator of the mobile device, and particular images may be selected for further analysis in various manners. Furthermore, the calculation of object volume and/or other determined object information may include generating and manipulating a computer model or other representation of the object from selected images.

Posted Content
TL;DR: Wang et al. as mentioned in this paper proposed to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network.
Abstract: 3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representations automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet -- a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks.

Journal ArticleDOI
TL;DR: It is shown that similar unisensory visual and haptic representations lead to a shared mult isensory representation underlying both cross-modal object recognition and view-independence, which suggests a common neural substrate.
Abstract: Visual and haptic unisensory object processing show many similarities in terms of categorization, recognition, and representation. In this review, we discuss how these similarities contribute to multisensory object processing. In particular, we show that similar unisensory visual and haptic representations lead to a shared multisensory representation underlying both cross-modal object recognition and view-independence. This shared representation suggests a common neural substrate and we review several candidate brain regions, previously thought to be specialized for aspects of visual processing, that are now known also to be involved in analogous haptic tasks. Finally, we lay out the evidence for a model of multisensory object recognition in which top-down and bottom-up pathways to the object-selective lateral occipital complex are modulated by object familiarity and individual differences in object and spatial imagery.

Journal ArticleDOI
TL;DR: In seven experiments, direct parallels are shown between spatial demonstrative usage in English and (non-linguistic) memory for object location, indicating close connections between the language of space and non-lingUistic spatial representation.

Proceedings ArticleDOI
29 Sep 2014
TL;DR: In this work, an approach is presented that estimates a uniform, low-level, grid-based world model including dynamic and static objects, their uncertainties, as well as their velocities, which does not require existing object tracks to filter out data points not used for creating and updating the map.
Abstract: Mapping and tracking in dynamic environments for autonomously-moving robots is still challenging, despite being essential tasks. They are often done separately using occupancy grids and established object tracking algorithms. In this work, an approach is presented that estimates a uniform, low-level, grid-based world model including dynamic and static objects, their uncertainties, as well as their velocities. It does not require existing object tracks to filter out data points not used for creating and updating the map. Nor does it require that measurements can be classified into belonging to a static or to a moving object. Promising results from experiments with an autonomous vehicle equipped with a laser scanner demonstrate the usefulness of the approach.

01 Jan 2014
TL;DR: In this article, the authors show that the integration of visual and motor-relevant object information occurs at the level of single OTC areas and provide evidence that the ventral visual pathway is actively and flexibly engaged in processing object weight, an object property critical for action planning and control.
Abstract: Skilled manipulation requires the ability to predict the weights of viewed objects based on learned associations linking object weight to object visual appearance. However, the neural mechanisms involved in extracting weight information from viewed object properties are unknown. Given that ventral visual pathway areas represent a wide variety of object features, one intriguing but as yet untested possibility is that these areas also represent object weight, a nonvisual motor-relevant object property. Here, using event-related fMRI and pattern classification techniques, we tested the novel hypothesis that object-sensitive regions in occipitotemporal cortex (OTC), in addition to traditional motor-related brain areas, represent object weight when preparing to lift that object. In two studies, the same participants prepared and then executed lifting actions with objects of varying weight. In the first study, we show that when lifting visually identical objects, where predicted weight is based solely on sensorimotor memory, weight is represented in object-sensitive OTC. In the second study, we show that when object weight is associated with a particular surface texture, that texture-sensitive OTC areas also come to represent object weight. Notably, these texture-sensitive areas failed to carry information about weight in the first study, when object surface properties did not specify weight. Our results indicate that the integration of visual and motor-relevant object information occurs at the level of single OTC areas and provide evidence that the ventral visual pathway is actively and flexibly engaged in processing object weight, an object property critical for action planning and control.

Patent
08 Jan 2014
TL;DR: In this paper, a method for receiving processed information at a remote device is described, which includes transmitting from the remote device a verbal request to a first information provider and receiving a digital message from the first information providers in response to the transmitted verbal request.
Abstract: A method for receiving processed information at a remote device is described. The method includes transmitting from the remote device a verbal request to a first information provider and receiving a digital message from the first information provider in response to the transmitted verbal request. The digital message includes a symbolic representation indicator associated with a symbolic representation of the verbal request and data used to control an application. The method also includes transmitting, using the application, the symbolic representation indicator to a second information provider for generating results to be displayed on the remote device.

Journal ArticleDOI
TL;DR: This review argues that the ability to interact with objects the authors cannot see implies an internal memory model of the surroundings, available to the motor system, and that this representation has a location in the precuneus, on the medial surface of the superior parietal cortex.
Abstract: Our phenomenal world remains stationary in spite of movements of the eyes, head and body. In addition, we can point or turn to objects in the surroundings whether or not they are in the field of view. In this review, I argue that these two features of experience and behaviour are related. The ability to interact with objects we cannot see implies an internal memory model of the surroundings, available to the motor system. And, because we maintain this ability when we move around, the model must be updated, so that the locations of object memories change continuously to provide accurate directional information. The model thus contains an internal representation of both the surroundings and the motions of the head and body: in other words, a stable representation of space. Recent functional MRI studies have provided strong evidence that this egocentric representation has a location in the precuneus, on the medial surface of the superior parietal cortex. This is a region previously identified with ‘self-centred mental imagery’, so it seems likely that the stable egocentric representation, required by the motor system, is also the source of our conscious percept of a stable world.

Journal ArticleDOI
TL;DR: It is argued that, although progress has been made, inequity in academic medicine matters because some areas of medicine are underresearched at a cost to patients and society.
Abstract: In 2008, the then Chief Medical Officer commissioned Baroness Deech to chair an Independent Working Group looking at the position and participation of women in the medical profession. We update and extend the Deech report to cover academic medicine in the UK. In doing so, we are not offering a systematic or exhaustive review of the data. Rather, we describe female participation rates in medicine and academic medicine based on secondary analysis. We demonstrate that although women are equally represented in medicine, they are under-represented in academic medicine. We conclude by arguing that, although progress has been made, inequity in academic medicine matters because: (a) it is a waste of public investment due to a loss of research talent; (b) as a consequence, some areas of medicine are underresearched at a cost to patients and society; and, (c) discriminatory practices and unconscious bias continue to occur.

BookDOI
05 Mar 2014
TL;DR: In this paper, the authors argue that while new tools and approaches for understanding cognition are valuable, representational approaches do not need to be abandoned in the course of constructing new models and explanations and that models that incorporate representation are quite compatible with the kinds of complex situations being modeled with the new methods.
Abstract: Recent work in cognitive science, much of it placed in opposition to a computational view of the mind, has argued that the concept of representation and theories based on that concept are not sufficient to explain the details of cognitive processing. These attacks on representation have focused on the importance of context sensitivity in cognitive processing, on the range of individual differences in performance, and on the relationship between minds and the bodies and environments in which they exist. In each case, models based on traditional assumptions about representation have been assumed to be too rigid to account for the effects of these factors on cognitive processing. In place of a representational view of mind, other formalisms and methodologies, such as nonlinear differential equations (or dynamical systems) and situated robotics, have been proposed as better explanatory tools for understanding cognition. This book is based on the notion that, while new tools and approaches for understanding cognition are valuable, representational approaches do not need to be abandoned in the course of constructing new models and explanations. Rather, models that incorporate representation are quite compatible with the kinds of complex situations being modeled with the new methods. This volume illustrates the power of this explicitly representational approach--labeled "cognitive dynamics"--in original essays by prominent researchers in cognitive science. Each chapter explores some aspect of the dynamics of cognitive processing while still retaining representations as the centerpiece of the explanations of the key phenomena. These chapters serve as an existence proof that representation is not incompatible with the dynamics of cognitive processing. The book is divided into sections on foundational issues about the use of representation in cognitive science, the dynamics of low level cognitive processes (such as visual and auditory perception and simple lexical priming), and the dynamics of higher cognitive processes (including categorization, analogy, and decision making).


Patent
15 Sep 2014
TL;DR: In this paper, a visual representation of the first data flow based on the elements relating to the data is generated, and the visual representation is adjusted according to the first tag in response to selection of the tag.
Abstract: Systems, methods, and non-transitory computer readable media configured to capture a first data flow between a data source and a data client. One or more elements relating to the first data flow are determined. At least one element of the first data flow is tagged with a first tag. A visual representation of the first data flow based on the elements relating to the data is generated. The visual representation of the first data flow is adjusted according to the first tag in response to selection of the first tag.