scholarly journals Category Boundaries and Typicality Warp the Neural Representation Space of Real-World Object Categories

2015 ◽  
Vol 15 (12) ◽  
pp. 8
Author(s):  
Marius Catalin Iordan ◽  
Michelle Greene ◽  
Diane Beck ◽  
Li Fei-Fei
Keyword(s):  
Real World ◽  
Neural Representation ◽  
Representation Space ◽  
Object Categories

scholarly journals The neural representation of dynamic real-world auditory/visual events

2010 ◽  
Vol 8 (6) ◽  
pp. 1054-1054
Author(s):  
J. Vettel ◽  
J. Green ◽  
L. Heller ◽  
M. Tarr
Keyword(s):  
Real World ◽  
Neural Representation ◽  
Visual Events

scholarly journals Neural Representation and Learning of Hierarchical 2-additive Choquet Integrals

2020 ◽  
Author(s):  
Roman Bresson ◽  
Johanne Cohen ◽  
Eyke Hüllermeier ◽  
Christophe Labreuche ◽  
Michèle Sebag
Keyword(s):  
Real World ◽  
Choquet Integral ◽  
Marginal Utility ◽  
Decision Makers ◽  
Neural Representation ◽  
Automatic Identification ◽  
Neural Architecture ◽  
Real World Applications ◽  
Choquet Integrals ◽  
The Cost

Multi-Criteria Decision Making (MCDM) aims at modelling expert preferences and assisting decision makers in identifying options best accommodating expert criteria. An instance of MCDM model, the Choquet integral is widely used in real-world applications, due to its ability to capture interactions between criteria while retaining interpretability. Aimed at a better scalability and modularity, hierarchical Choquet integrals involve intermediate aggregations of the interacting criteria, at the cost of a more complex elicitation. The paper presents a machine learning-based approach for the automatic identification of hierarchical MCDM models, composed of 2-additive Choquet integral aggregators and of marginal utility functions on the raw features from data reflecting expert preferences. The proposed NEUR-HCI framework relies on a specific neural architecture, enforcing by design the Choquet model constraints and supporting its end-to-end training. The empirical validation of NEUR-HCI on real-world and artificial benchmarks demonstrates the merits of the approach compared to state-of-art baselines.

scholarly journals Real-world visual statistics and infants' first-learned object names

2017 ◽  
Vol 372 (1711) ◽  
pp. 20160055 ◽  
Author(s):  
Elizabeth M. Clerkin ◽  
Elizabeth Hart ◽  
James M. Rehg ◽  
Chen Yu ◽  
Linda B. Smith
Keyword(s):  
Statistical Learning ◽  
Frequency Distribution ◽  
Word Learning ◽  
Real World ◽  
Computational Models ◽  
Unsolved Problem ◽  
Statistical Structure ◽  
Cognitive Sciences ◽  
Object Categories ◽  
Small Set

We offer a new solution to the unsolved problem of how infants break into word learning based on the visual statistics of everyday infant-perspective scenes. Images from head camera video captured by 8 1/2 to 10 1/2 month-old infants at 147 at-home mealtime events were analysed for the objects in view. The images were found to be highly cluttered with many different objects in view. However, the frequency distribution of object categories was extremely right skewed such that a very small set of objects was pervasively present—a fact that may substantially reduce the problem of referential ambiguity. The statistical structure of objects in these infant egocentric scenes differs markedly from that in the training sets used in computational models and in experiments on statistical word-referent learning. Therefore, the results also indicate a need to re-examine current explanations of how infants break into word learning. This article is part of the themed issue ‘New frontiers for statistical learning in the cognitive sciences’.

Neural Correlates of Divided Attention in Natural Scenes

2016 ◽  
Vol 28 (9) ◽  
pp. 1392-1405 ◽  
Author(s):  
Sabrina Fagioli ◽  
Emiliano Macaluso
Keyword(s):  
Real World ◽  
Divided Attention ◽  
Premotor Cortex ◽  
Natural Scenes ◽  
Multiple Streams ◽  
Top Down ◽  
Frontoparietal Network ◽  
Object Categories ◽  
Visual Hemifields ◽  
Spatial Locations

Individuals are able to split attention between separate locations, but divided spatial attention incurs the additional requirement of monitoring multiple streams of information. Here, we investigated divided attention using photos of natural scenes, where the rapid categorization of familiar objects and prior knowledge about the likely positions of objects in the real world might affect the interplay between these spatial and nonspatial factors. Sixteen participants underwent fMRI during an object detection task. They were presented with scenes containing either a person or a car, located on the left or right side of the photo. Participants monitored either one or both object categories, in one or both visual hemifields. First, we investigated the interplay between spatial and nonspatial attention by comparing conditions of divided attention between categories and/or locations. We then assessed the contribution of top–down processes versus stimulus-driven signals by separately testing the effects of divided attention in target and nontarget trials. The results revealed activation of a bilateral frontoparietal network when dividing attention between the two object categories versus attending to a single category but no main effect of dividing attention between spatial locations. Within this network, the left dorsal premotor cortex and the left intraparietal sulcus were found to combine task- and stimulus-related signals. These regions showed maximal activation when participants monitored two categories at spatially separate locations and the scene included a nontarget object. We conclude that the dorsal frontoparietal cortex integrates top–down and bottom–up signals in the presence of distractors during divided attention in real-world scenes.

Tracking the Neural Representation of Context and Emotion During Reminiscence of Real-World Experiences

2012 ◽  
Author(s):  
Troy A. Smith ◽  
William A. Cunningham ◽  
Simon Dennis ◽  
Per B. Sederberg
Keyword(s):  
Real World ◽  
Neural Representation

scholarly journals Cortical Networks Representing Object Categories and High-level Attributes of Familiar Real-world Action Sounds

2011 ◽  
Vol 23 (8) ◽  
pp. 2079-2101 ◽  
Author(s):  
James W. Lewis ◽  
William J. Talkington ◽  
Aina Puce ◽  
Lauren R. Engel ◽  
Chris Frum
Keyword(s):  
Sound Source ◽  
Real World ◽  
Visual Object ◽  
Object Processing ◽  
Object Categories ◽  
Before And After ◽  
Listening Experience ◽  
Cortical Regions ◽  
The Stability ◽  
High Level

In contrast to visual object processing, relatively little is known about how the human brain processes everyday real-world sounds, transforming highly complex acoustic signals into representations of meaningful events or auditory objects. We recently reported a fourfold cortical dissociation for representing action (nonvocalization) sounds correctly categorized as having been produced by human, animal, mechanical, or environmental sources. However, it was unclear how consistent those network representations were across individuals, given potential differences between each participant's degree of familiarity with the studied sounds. Moreover, it was unclear what, if any, auditory perceptual attributes might further distinguish the four conceptual sound-source categories, potentially revealing what might drive the cortical network organization for representing acoustic knowledge. Here, we used functional magnetic resonance imaging to test participants before and after extensive listening experience with action sounds, and tested for cortices that might be sensitive to each of three different high-level perceptual attributes relating to how a listener associates or interacts with the sound source. These included the sound's perceived concreteness, effectuality (ability to be affected by the listener), and spatial scale. Despite some variation of networks for environmental sounds, our results verified the stability of a fourfold dissociation of category-specific networks for real-world action sounds both before and after familiarity training. Additionally, we identified cortical regions parametrically modulated by each of the three high-level perceptual sound attributes. We propose that these attributes contribute to the network-level encoding of category-specific acoustic knowledge representations.

scholarly journals Multivariate sensitivity to voice during auditory categorization

2015 ◽  
Vol 114 (3) ◽  
pp. 1819-1826 ◽  
Author(s):  
Yune Sang Lee ◽  
Jonathan E. Peelle ◽  
David Kraemer ◽  
Samuel Lloyd ◽  
Richard Granger
Keyword(s):  
Brain Activity ◽  
Temporal Cortex ◽  
Neural Representation ◽  
Current Evidence ◽  
General Role ◽  
Object Categories ◽  
Novel Approach ◽  
Auditory Object ◽  
Mri Study ◽  
Multivariate Pattern

Past neuroimaging studies have documented discrete regions of human temporal cortex that are more strongly activated by conspecific voice sounds than by nonvoice sounds. However, the mechanisms underlying this voice sensitivity remain unclear. In the present functional MRI study, we took a novel approach to examining voice sensitivity, in which we applied a signal detection paradigm to the assessment of multivariate pattern classification among several living and nonliving categories of auditory stimuli. Within this framework, voice sensitivity can be interpreted as a distinct neural representation of brain activity that correctly distinguishes human vocalizations from other auditory object categories. Across a series of auditory categorization tests, we found that bilateral superior and middle temporal cortex consistently exhibited robust sensitivity to human vocal sounds. Although the strongest categorization was in distinguishing human voice from other categories, subsets of these regions were also able to distinguish reliably between nonhuman categories, suggesting a general role in auditory object categorization. Our findings complement the current evidence of cortical sensitivity to human vocal sounds by revealing that the greatest sensitivity during categorization tasks is devoted to distinguishing voice from nonvoice categories within human temporal cortex.

scholarly journals Disorder and the neural representation of complex odors: smelling in the real world

2017 ◽  
Author(s):  
Kamesh Krishnamurthy ◽  
Ann M. Hermundstad ◽  
Thierry Mora ◽  
Aleksandra M. Walczak ◽  
Vijay Balasubramanian
Keyword(s):  
Real World ◽  
Neural Representation ◽  
Vast Number ◽  
The Real ◽  
Central Brain ◽  
Robustness To Noise ◽  
Molecular Mixtures ◽  
New Interpretation ◽  
Using Data ◽  
Odor Space

Animals smelling in the real world use a small number of receptors to sense a vast number of natural molecular mixtures, and proceed to learn arbitrary associations between odors and valences. Here, we propose a new interpretation of how the architecture of olfactory circuits is adapted to meet these immense complementary challenges. First, the diffuse binding of receptors to many molecules compresses a vast odor space into a tiny receptor space, while preserving similarity. Next, lateral interactions “densify” and decorrelate the response, enhancing robustness to noise. Finally, disordered projections from the periphery to the central brain reconfigure the densely packed information into a format suitable for flexible learning of associations and valences. We test our theory empirically using data from Drosophila. Our theory suggests that the neural processing of olfactory information differs from the other senses in its fundamental use of disorder.

Sign in / Sign up

Export Citation Format

Share Document