scholarly journals New End-Stopped Complex Cell Model Applicable in Convolutional Neural Networks

2021 ◽  
Vol 19 (3) ◽  
pp. 50-60
Author(s):  
A. V. Kugaevskikh
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Gabor Filter ◽  
Cell Model ◽  
Complex Cell ◽  
Simple Cells ◽  
Complex Cells ◽  
Accurate Model ◽  
Proposed Model ◽  
Simple And Complex Cells

This article is dedicated to modeling the end-stopped neuron. This type of neuron gives the maximum response at the end of the line and is used to refine the edge. The article provides an overview of different models of end-stopped neurons. I have proposed a simpler and more accurate model of an end-stopped neuron based on the use of Gabor filters in antiphase. For this purpose, the models of simple and complex cells whose output is used in the proposed model are also described. Simple cells are based on the use of a Gabor filter, the parameters of which are also described in this article. The proposed model has shown its effectiveness.

Computing relative motion with complex cells

2005 ◽  
Vol 22 (2) ◽  
pp. 225-236 ◽  
Author(s):  
BABETTE K. DELLEN ◽  
JOHN W. CLARK ◽  
RALF WESSEL
Keyword(s):  
Relative Motion ◽  
Cell Model ◽  
Neural Representation ◽  
Complex Cell ◽  
Contextual Influences ◽  
Complex Cells ◽  
Cell Responses ◽  
Motion Computation ◽  
Model Tuning ◽  
Simple And Complex Cells

Contextual influences shape our perception of local visual stimuli. Relative-motion stimuli represent an important contextual influence, yet the mechanism subserving relative-motion computation remains largely unknown. In the present work, we investigated the responses of an established model for simple and complex cells to relative-motion stimuli. A straightforward mathematical analysis showed that relative-motion computation is inherent in the nonlinear transformation of the complex-cell model. Tuning to relative velocity is achieved by applying a temporal filter to the complex-cell response. The mathematical inference is supported by simulations that quantitatively reproduce measured complex-cell responses in both cat and monkey to a variety of relative-motion stimuli. Importantly, the posited mechanism for cortical computation of relative motion does not require an intermediate neural representation of local velocities and does not require lateral or feedback interactions within a network.

scholarly journals Stock Pattern Classification from Charts using Deep Learning Algorithms

2020 ◽  
Vol 3 (1) ◽  
pp. 445-454
Author(s):  
Celal Buğra Kaya ◽  
Alperen Yılmaz ◽  
Gizem Nur Uzun ◽  
Zeynep Hilal Kilimci
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Convolutional Neural Networks ◽  
Pattern Classification ◽  
Short Term Memory ◽  
Stock Exchange ◽  
Learning Techniques ◽  
Proposed Model ◽  
Istanbul Stock Exchange

Pattern classification is related with the automatic finding of regularities in dataset through the utilization of various learning techniques. Thus, the classification of the objects into a set of categories or classes is provided. This study is undertaken to evaluate deep learning methodologies to the classification of stock patterns. In order to classify patterns that are obtained from stock charts, convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long-short term memory networks (LSTMs) are employed. To demonstrate the efficiency of proposed model in categorizing patterns, hand-crafted image dataset is constructed from stock charts in Istanbul Stock Exchange and NASDAQ Stock Exchange. Experimental results show that the usage of convolutional neural networks exhibits superior classification success in recognizing patterns compared to the other deep learning methodologies.

Direction selectivity of complex cells in a comparison with simple cells

1975 ◽  
Vol 38 (6) ◽  
pp. 1524-1540 ◽  
Author(s):  
A. W. Goodwin ◽  
G. H. Henry
Keyword(s):  
Spontaneous Activity ◽  
Visual Information ◽  
Striate Cortex ◽  
Cell Types ◽  
Direction Selectivity ◽  
Complex Cell ◽  
Simple Cells ◽  
Complex Cells ◽  
Sequential Processing

Following our earlier study on direction selectivity in simple cells (5), the present findings on complex cells made it possible to compare the direction selectivity in the two types of striate cell. Common properties were found in the dimension of the smallest stimulus displacement giving a direction-selective response and in the role of inhibition in suppressing the response as the stimulus moved in the nonpreferred direction. However, the effectiveness of this inhibition varied in the two cell types since it suppressed both driven and spontaneous activity in the simple cell, but only driven firing in the complex cell. It is argued that direction selectivity must enter the response before the complex cell if the inhibition responsible for it's generation fails to influence the spontaneous activity of the cell. The consequences of this finding are considered in the terms of parallel or sequential processing of visual information in striate cortex.

scholarly journals Auditory Inspired Convolutional Neural Networks for Ship Type Classification with Raw Hydrophone Data

Entropy ◽  
2018 ◽  
Vol 20 (12) ◽  
pp. 990 ◽  
Author(s):  
Sheng Shen ◽  
Honghui Yang ◽  
Junhao Li ◽  
Guanghui Xu ◽  
Meiping Sheng
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Filter Banks ◽  
Auditory Filter ◽  
Feature Representations ◽  
Proposed Model ◽  
Energy Pooling ◽  
Practical Guidelines ◽  
Frequency Components ◽  
Underwater Acoustic Signal

Detecting and classifying ships based on radiated noise provide practical guidelines for the reduction of underwater noise footprint of shipping. In this paper, the detection and classification are implemented by auditory inspired convolutional neural networks trained from raw underwater acoustic signal. The proposed model includes three parts. The first part is performed by a multi-scale 1D time convolutional layer initialized by auditory filter banks. Signals are decomposed into frequency components by convolution operation. In the second part, the decomposed signals are converted into frequency domain by permute layer and energy pooling layer to form frequency distribution in auditory cortex. Then, 2D frequency convolutional layers are applied to discover spectro-temporal patterns, as well as preserve locality and reduce spectral variations in ship noise. In the third part, the whole model is optimized with an objective function of classification to obtain appropriate auditory filters and feature representations that are correlative with ship categories. The optimization reflects the plasticity of auditory system. Experiments on five ship types and background noise show that the proposed approach achieved an overall classification accuracy of 79.2%, which improved by 6% compared to conventional approaches. Auditory filter banks were adaptive in shape to improve accuracy of classification.

scholarly journals BE-FNet: 3D Bounding Box Estimation Feature Pyramid Network for Accurate and Efficient Maxillary Sinus Segmentation

2020 ◽  
Vol 2020 ◽  
pp. 1-16
Author(s):  
Zhuofu Deng ◽  
Binbin Wang ◽  
Zhiliang Zhu
Keyword(s):  
Neural Networks ◽  
Maxillary Sinus ◽  
Convolutional Neural Networks ◽  
Feature Fusion ◽  
Medical Images ◽  
Medical Image Segmentation ◽  
3D Segmentation ◽  
Deep Convolutional Neural Networks ◽  
Bounding Box ◽  
Proposed Model

Maxillary sinus segmentation plays an important role in the choice of therapeutic strategies for nasal disease and treatment monitoring. Difficulties in traditional approaches deal with extremely heterogeneous intensity caused by lesions, abnormal anatomy structures, and blurring boundaries of cavity. 2D and 3D deep convolutional neural networks have grown popular in medical image segmentation due to utilization of large labeled datasets to learn discriminative features. However, for 3D segmentation in medical images, 2D networks are not competent in extracting more significant spacial features, and 3D ones suffer from unbearable burden of computation, which results in great challenges to maxillary sinus segmentation. In this paper, we propose a deep neural network with an end-to-end manner to generalize a fully automatic 3D segmentation. At first, our proposed model serves a symmetrical encoder-decoder architecture for multitask of bounding box estimation and in-region 3D segmentation, which cannot reduce excessive computation requirements but eliminate false positives remarkably, promoting 3D segmentation applied in 3D convolutional neural networks. In addition, an overestimation strategy is presented to avoid overfitting phenomena in conventional multitask networks. Meanwhile, we introduce residual dense blocks to increase the depth of the proposed network and attention excitation mechanism to improve the performance of bounding box estimation, both of which bring little influence to computation cost. Especially, the structure of multilevel feature fusion in the pyramid network strengthens the ability of identification to global and local discriminative features in foreground and background achieving more advanced segmentation results. At last, to address problems of blurring boundary and class imbalance in medical images, a hybrid loss function is designed for multiple tasks. To illustrate the strength of our proposed model, we evaluated it against the state-of-the-art methods. Our model performed better significantly with an average Dice 0.947±0.031, VOE 10.23±5.29, and ASD 2.86±2.11, respectively, which denotes a promising technique with strong robust in practice.

scholarly journals Spatial and Temporal Features of Synaptic to Discharge Receptive Field Transformation in Cat Area 17

2010 ◽  
Vol 103 (2) ◽  
pp. 677-697 ◽  
Author(s):  
Lionel G. Nowak ◽  
Maria V. Sanchez-Vives ◽  
David A. McCormick
Keyword(s):  
Spatial Frequency ◽  
Response Latency ◽  
Frequency Tuning ◽  
Receptive Fields ◽  
Width Ratio ◽  
Area 17 ◽  
Simple Cells ◽  
Complex Cells ◽  
Temporal Features ◽  
Simple And Complex Cells

The aim of the present study was to characterize the spatial and temporal features of synaptic and discharge receptive fields (RFs), and to quantify their relationships, in cat area 17. For this purpose, neurons were recorded intracellularly while high-frequency flashing bars were used to generate RFs maps for synaptic and spiking responses. Comparison of the maps shows that some features of the discharge RFs depended strongly on those of the synaptic RFs, whereas others were less dependent. Spiking RF duration depended poorly and spiking RF amplitude depended moderately on those of the underlying synaptic RFs. At the other extreme, the optimal spatial frequency and phase of the discharge RFs in simple cells were almost entirely inherited from those of the synaptic RFs. Subfield width, in both simple and complex cells, was less for spiking responses compared with synaptic responses, but synaptic to discharge width ratio was relatively variable from cell to cell. When considering the whole RF of simple cells, additional variability in width ratio resulted from the presence of additional synaptic subfields that remained subthreshold. Due to these additional, subthreshold subfields, spatial frequency tuning predicted from synaptic RFs appears sharper than that predicted from spiking RFs. Excitatory subfield overlap in spiking RFs was well predicted by subfield overlap at the synaptic level. When examined in different regions of the RF, latencies appeared to be quite variable, but this variability showed negligible dependence on distance from the RF center. Nevertheless, spiking response latency faithfully reflected synaptic response latency.

scholarly journals Deep Learning for Detecting Building Defects Using Convolutional Neural Networks

Sensors ◽  
2019 ◽  
Vol 19 (16) ◽  
pp. 3556 ◽  
Author(s):  
Husein Perez ◽  
Joseph H. M. Tah ◽  
Amir Mosavi
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Effective Means ◽  
Scale Up ◽  
Real Life ◽  
Health And Safety ◽  
Condition Assessment ◽  
Automated Detection ◽  
Proposed Model ◽  
Current Asset

Clients are increasingly looking for fast and effective means to quickly and frequently survey and communicate the condition of their buildings so that essential repairs and maintenance work can be done in a proactive and timely manner before it becomes too dangerous and expensive. Traditional methods for this type of work commonly comprise of engaging building surveyors to undertake a condition assessment which involves a lengthy site inspection to produce a systematic recording of the physical condition of the building elements, including cost estimates of immediate and projected long-term costs of renewal, repair and maintenance of the building. Current asset condition assessment procedures are extensively time consuming, laborious, and expensive and pose health and safety threats to surveyors, particularly at height and roof levels which are difficult to access. This paper aims at evaluating the application of convolutional neural networks (CNN) towards an automated detection and localisation of key building defects, e.g., mould, deterioration, and stain, from images. The proposed model is based on pre-trained CNN classifier of VGG-16 (later compaired with ResNet-50, and Inception models), with class activation mapping (CAM) for object localisation. The challenges and limitations of the model in real-life applications have been identified. The proposed model has proven to be robust and able to accurately detect and localise building defects. The approach is being developed with the potential to scale-up and further advance to support automated detection of defects and deterioration of buildings in real-time using mobile devices and drones.

scholarly journals Substructure of Direction-Selective Receptive Fields in Macaque V1

2003 ◽  
Vol 89 (5) ◽  
pp. 2743-2759 ◽  
Author(s):  
Margaret S. Livingstone ◽  
Bevil R. Conway
Keyword(s):  
Receptive Fields ◽  
Direction Selectivity ◽  
Positive Interactions ◽  
Null Direction ◽  
Simple Cells ◽  
Complex Cells ◽  
Preferred Direction ◽  
Orientation Axis ◽  
Interaction Regions ◽  
Simple And Complex Cells

We used two-dimensional (2-D) sparse noise to map simultaneous and sequential two-spot interactions in simple and complex direction-selective cells in macaque V1. Sequential-interaction maps for both simple and complex cells showed preferred-direction facilitation and null-direction suppression for same-contrast stimulus sequences and the reverse for inverting-contrast sequences, although the magnitudes of the interactions were weaker for the simple cells. Contrast-sign selectivity in complex cells indicates that direction-selective interactions in these cells must occur in antecedent simple cells or in simple-cell-like dendritic compartments. Our maps suggest that direction selectivity, and on andoff segregation perpendicular to the orientation axis, can occur prior to receptive-field elongation along the orientation axis. 2-D interaction maps for some complex cells showed elongated alternating facilitatory and suppressive interactions as predicted if their inputs were orientation-selective simple cells. The negative interactions, however, were less elongated than the positive interactions, and there was an inflection at the origin in the positive interactions, so the interactions were chevron-shaped rather than band-like. Other complex cells showed only two round interaction regions, one negative and one positive. Several explanations for the map shapes are considered, including the possibility that directional interactions are generated directly from unoriented inputs.

Sign in / Sign up

Export Citation Format

Share Document