scholarly journals Your favorite color makes learning more precise and adaptable

2017 ◽  
Author(s):  
Shiva Farashahi ◽  
Katherine Rowe ◽  
Zohra Aslami ◽  
Daeyeol Lee ◽  
Alireza Soltani
Keyword(s):  
Computational Models ◽  
Human Subjects ◽  
Dynamic Environments ◽  
Parallel Learning ◽  
Feature Based ◽  
Favorite Color ◽  
Individual Features ◽  
Feature Values ◽  
Low Dimensional

AbstractLearning from reward feedback is essential for survival but can become extremely challenging with myriad choice options. Here, we propose that learning reward values of individual features can provide a heuristic for estimating reward values of choice options in dynamic, multidimensional environments. We hypothesized that this feature-based learning occurs not just because it can reduce dimensionality, but more importantly because it can increase adaptability without compromising precision of learning. We experimentally tested this hypothesis and found that in dynamic environments, human subjects adopted feature-based learning even when this approach does not reduce dimensionality. Even in static, low-dimensional environments, subjects initially adopted feature-based learning and gradually switched to learning reward values of individual options, depending on how accurately objects’ values can be predicted by combining feature values. Our computational models reproduced these results and highlight the importance of neurons coding feature values for parallel learning of values for features and objects.

scholarly journals Feature-based data assimilation in geophysics

2017 ◽  
Author(s):  
Matthias Morzfeld ◽  
Jesse Adams ◽  
Spencer Lunderman ◽  
Rafael Orozco
Keyword(s):  
Computational Models ◽  
Chaotic Systems ◽  
Efficient Solutions ◽  
Effective Dimension ◽  
Dimensional Model ◽  
Computationally Efficient ◽  
Fundamental Understanding ◽  
Feature Based ◽  
Low Dimensional ◽  
Noise Models

Abstract. Many applications in science require that computational models and data be combined. In a Bayesian framework, this is usually done by defining likelihoods based on the mismatch of model outputs and data. However, matching model outputs and data in this way can be unnecessary or impossible. For example, using large amounts of steady state data is unnecessary because these data are redundant, it is numerically difficult to assimilate data in chaotic systems, and it is often impossible to assimilate data of a complex system into a low-dimensional model. These issues can be addressed by selecting features of the data, and defining likelihoods based on the features, rather than by the usual mismatch of model output and data. Our goal is to contribute to a fundamental understanding of such a feature-based approach that allows us to assimilate selected aspects of data into models. Specifically, we explain how the feature-based approach can be interpreted as a method for reducing an effective dimension, and derive new noise models, based on perturbed observations, that lead to computationally efficient solutions. Numerical implementations of our ideas are illustrated in four examples.

scholarly journals THE ROLE OF GLOBAL AND FEATURE BASED INFORMATION IN GENDER CLASSIFICATION OF FACES: A COMPARISON OF HUMAN PERFORMANCE AND COMPUTATIONAL MODELS

2005 ◽  
Vol 15 (01n02) ◽  
pp. 121-128 ◽  
Author(s):  
SAMARASENA BUCHALA ◽  
NEIL DAVEY ◽  
RAY J. FRANK ◽  
MARTIN LOOMES ◽  
TIM M. GALE
Keyword(s):  
Dimensionality Reduction ◽  
Computational Models ◽  
Human Performance ◽  
Human Subjects ◽  
Support Vector ◽  
Equal Weight ◽  
Gender Classification ◽  
Reduction Techniques ◽  
Dimensionality Reduction Techniques ◽  
Feature Based

Most computational models for gender classification use global information (the full face image) giving equal weight to the whole face area irrespective of the importance of the internal features. Here, we use a global and feature based representation of face images that includes both global and featural information. We use dimensionality reduction techniques and a support vector machine classifier and show that this method performs better than either global or feature based representations alone. We also present results of human subjects performance on gender classification task and evaluate how the different dimensionality reduction techniques compare with human subjects performance. The results support the psychological plausibility of the global and feature based representation.

scholarly journals Feature-based data assimilation in geophysics

2018 ◽  
Vol 25 (2) ◽  
pp. 355-374 ◽  
Author(s):  
Matthias Morzfeld ◽  
Jesse Adams ◽  
Spencer Lunderman ◽  
Rafael Orozco
Keyword(s):  
Computational Models ◽  
Chaotic Systems ◽  
Efficient Solutions ◽  
Effective Dimension ◽  
Dimensional Model ◽  
Computationally Efficient ◽  
Fundamental Understanding ◽  
Feature Based ◽  
Low Dimensional ◽  
Noise Models

Abstract. Many applications in science require that computational models and data be combined. In a Bayesian framework, this is usually done by defining likelihoods based on the mismatch of model outputs and data. However, matching model outputs and data in this way can be unnecessary or impossible. For example, using large amounts of steady state data is unnecessary because these data are redundant. It is numerically difficult to assimilate data in chaotic systems. It is often impossible to assimilate data of a complex system into a low-dimensional model. As a specific example, consider a low-dimensional stochastic model for the dipole of the Earth's magnetic field, while other field components are ignored in the model. The above issues can be addressed by selecting features of the data, and defining likelihoods based on the features, rather than by the usual mismatch of model output and data. Our goal is to contribute to a fundamental understanding of such a feature-based approach that allows us to assimilate selected aspects of data into models. We also explain how the feature-based approach can be interpreted as a method for reducing an effective dimension and derive new noise models, based on perturbed observations, that lead to computationally efficient solutions. Numerical implementations of our ideas are illustrated in four examples.

scholarly journals Model-based fMRI reveals dissimilarity processes underlying base rate neglect

eLife ◽  
2018 ◽  
Vol 7 ◽  
Author(s):  
Sean R O'Bryan ◽  
Darrell A Worthy ◽  
Evan J Livesey ◽  
Tyler Davis
Keyword(s):  
Computational Models ◽  
Human Subjects ◽  
Base Rate ◽  
Imaging Data ◽  
Multivoxel Pattern Analysis ◽  
Activation Patterns ◽  
Model Based ◽  
Base Rate Neglect ◽  
Rate Information ◽  
Rare Category

Extensive evidence suggests that people use base rate information inconsistently in decision making. A classic example is the inverse base rate effect (IBRE), whereby participants classify ambiguous stimuli sharing features of both common and rare categories as members of the rare category. Computational models of the IBRE have either posited that it arises from associative similarity-based mechanisms or dissimilarity-based processes that may depend upon higher-level inference. Here we develop a hybrid model, which posits that similarity- and dissimilarity-based evidence both contribute to the IBRE, and test it using functional magnetic resonance imaging data collected from human subjects completing an IBRE task. Consistent with our model, multivoxel pattern analysis reveals that activation patterns on ambiguous test trials contain information consistent with dissimilarity-based processing. Further, trial-by-trial activation in left rostrolateral prefrontal cortex tracks model-based predictions for dissimilarity-based processing, consistent with theories positing a role for higher-level symbolic processing in the IBRE.

scholarly journals An approach to automatic boundary segmentation of solid models using virtual topology: toward reconstruction of design features

2020 ◽  
Vol 7 (3) ◽  
pp. 367-385
Author(s):  
Yingzhong Zhang ◽  
Yufei Fu ◽  
Jia Jia ◽  
Xiaofang Luo
Keyword(s):  
Design Feature ◽  
Boundary Representation ◽  
Virtual Topology ◽  
Design Features ◽  
Complex Part ◽  
Novel Approach ◽  
Solid Models ◽  
Medium Level ◽  
Feature Based ◽  
Low Dimensional

Abstract Boundary segmentation of solid models is the geometric foundation to reconstruct design features. In this paper, based on the shape evolution analysis for the feature-based modeling process, a novel approach to the automatic boundary segmentation of solid models for reconstructing design features is proposed. The presented approach simulates the designer’s decomposing thinking on how to decompose an existing boundary representation model into a set of design features. First, the modeling traces of design features are formally represented as a set of feature vertex adjacent graphs that use low-dimensional vertex entities and their connection relations. Then, a set of Boolean segmentation loops is searched and extracted from the constructed feature vertex adjacent graphs, which segment the boundary of a solid model into a set of regions with different design feature semantics. In the search process, virtual topology operations are employed to simulate the topological changes resulting from Boolean operations in feature modeling processes. In addition, to realize effective search, search strategies and search algorithms are presented. The segmentation experiments and case study show that the presented approach is feasible and effective for the boundary segmentation of medium-level complex part models. The presented approach lays the foundation for the later reconstruction of design features.

Low dimensional DCT and DWT feature based model for detection of image splicing and copy-move forgery

2020 ◽  
Vol 79 (39-40) ◽  
pp. 29977-30005 ◽  
Author(s):  
Sahani Pooja Jaiprakash ◽  
Madhavi B. Desai ◽  
Choudhary Shyam Prakash ◽  
Vipul H. Mistry ◽  
Kishankumar Lalajibhai Radadiya
Keyword(s):  
Image Splicing ◽  
Feature Based ◽  
Low Dimensional

scholarly journals Intrinsic motivation and episodic memories for robot exploration of high-dimensional sensory spaces

2020 ◽  
pp. 105971232092291
Author(s):  
Guido Schillaci ◽  
Antonio Pico Villalpando ◽  
Verena V Hafner ◽  
Peter Hanappe ◽  
David Colliaux ◽  
...  
Keyword(s):  
Neural Networks ◽  
Episodic Memory ◽  
Intrinsic Motivation ◽  
Computational Models ◽  
Deep Neural Networks ◽  
Image Sensor ◽  
Forward Kinematics ◽  
High Dimensional ◽  
Episodic Memories ◽  
Low Dimensional

This work presents an architecture that generates curiosity-driven goal-directed exploration behaviours for an image sensor of a microfarming robot. A combination of deep neural networks for offline unsupervised learning of low-dimensional features from images and of online learning of shallow neural networks representing the inverse and forward kinematics of the system have been used. The artificial curiosity system assigns interest values to a set of pre-defined goals and drives the exploration towards those that are expected to maximise the learning progress. We propose the integration of an episodic memory in intrinsic motivation systems to face catastrophic forgetting issues, typically experienced when performing online updates of artificial neural networks. Our results show that adopting an episodic memory system not only prevents the computational models from quickly forgetting knowledge that has been previously acquired but also provides new avenues for modulating the balance between plasticity and stability of the models.

scholarly journals High Dimensional Feature for Hyperspectral Image Classification

2018 ◽  
Vol 246 ◽  
pp. 03041
Author(s):  
Cailing Wang ◽  
Hongwei Wang ◽  
Yinyong Zhang ◽  
Jia Wen ◽  
Fan Yang
Keyword(s):  
Image Classification ◽  
Hyperspectral Image ◽  
Texture Feature ◽  
High Dimensional ◽  
Support Vector ◽  
Hyperspectral Image Classification ◽  
Feature Based ◽  
Pca Method ◽  
The One ◽  
Low Dimensional

Making a high dimensional (e.g., 100k-dim) feature for hyperspectral image classification seems not a good idea because it will bring difficulties on consequent training, computation, and storage. In this paper, we study the performance of a high-dimensional feature by texture feature. The texture feature based on multi-local binary pattern descriptor, can achieve significant improvements over both its tradition version and the one we proposed in our previous work. We also make the high-dimensional feature practical, we employ the PCA method for dimension reduction and support vector machine for hyperspectral image classification. The two real hyperspectral image datasets are employed. Our experimental results with real hyperspectral images indicate that the high dimensional feature can enhance the classification accuracy than some low dimensional.

Novel Visual Illusions Related to Vasarely's ‘Nested Squares’ Show That Corner Salience Varies with Corner Angle

Perception ◽  
10.1068/p5383 ◽  
2005 ◽  
Vol 34 (4) ◽  
pp. 409-420 ◽  
Author(s):  
Xoana G Troncoso ◽  
Stephen L Macknik ◽  
Susana Martinez-Conde
Keyword(s):  
Discrimination Task ◽  
Computational Models ◽  
Visual Illusion ◽  
Human Subjects ◽  
Brightness Discrimination ◽  
Receptive Fields ◽  
Visual Illusions ◽  
Corner Angle ◽  
Brightness Discrimination Task ◽  
Sharp Corners

Vasarely's ‘nested-squares’ illusion shows that 90° corners can be more salient perceptually than straight edges. On the basis of this illusion we have developed a novel visual illusion, the ‘Alternating Brightness Star’, which shows that sharp corners are more salient than shallow corners (an effect we call ‘corner angle salience variation’) and that the same corner can be perceived as either bright or dark depending on the polarity of the angle (ie whether concave or convex: ‘corner angle brightness reversal’). Here we quantify the perception of corner angle salience variation and corner angle brightness reversal effects in twelve naive human subjects, in a two-alternative forced-choice brightness discrimination task. The results show that sharp corners generate stronger percepts than shallow corners, and that corner gradients appear bright or dark depending on whether the corner is concave or convex. Basic computational models of center – surround receptive fields predict the results to some degree, but not fully.

scholarly journals Gaze-evoked deformations of the optic nerve head in thyroid eye disease

2020 ◽  
Author(s):  
Liam K. Fisher ◽  
Xiaofei Wang ◽  
Tin A. Tun ◽  
Hsi-Wei Chung ◽  
Dan Milea ◽  
...  
Keyword(s):  
Eye Movements ◽  
Optic Nerve ◽  
Optic Nerve Head ◽  
Tilt Angle ◽  
Computational Models ◽  
Human Subjects ◽  
Computational Modelling ◽  
Thyroid Eye Disease ◽  
Eye Disease ◽  
Angle Measurements

AbstractPurposeTo assess gaze evoked deformations of the optic nerve head (ONH) in thyroid eye disease (TED), using computational modelling and optical coherence tomography (OCT).MethodsMultiple finite element models were constructed: One model of a healthy eye, and two models mimicking effects of TED; one with proptosis and another with extraocular tissue stiffening. Two additional hypothetical models had extraocular tissue softening or no extraocular tissue at all. Horizontal eye movements were simulated in these models.OCT images of the ONH of 10 healthy volunteers and 1 patient with TED were taken in primary gaze. Additional images were recorded in the same subjects performing eye movements in adduction and abduction.The resulting ONH deformation in the models and human subjects was measured by recording the ‘tilt angle’ (relative antero-posterior deformation of the Bruch’s membrane opening). Effective stress was measured in the peripapillary sclera of the models.ResultsIn our computational models the eyes with proptosis and stiffer extraocular tissue had greater gaze-evoked deformations than the healthy eye model, while the models with softer or no extraocular tissue had lesser deformations, in both adduction and abduction. Scleral stress correlated with the tilt angle measurements.In healthy subjects, the mean tilt angle was 1.46° ± 0.25 in adduction and −0.42° ± 0.12 in abduction. The tilt angle measured in the subject with TED was 5.37° in adduction and −2.21° in abduction.ConclusionsComputational modelling and experimental observation suggest that TED can cause increased gaze-evoked deformations of the ONH.

Sign in / Sign up

Export Citation Format

Share Document