Neural Architecture Search for Lightweight Neural Network in Food Recognition

Healthy eating is an essential element to prevent obesity that will lead to chronic diseases. Despite numerous efforts to promote the awareness of healthy food consumption, the obesity rate has been increased in the past few years. An automated food recognition system is needed to serve as a fundamental source of information for promoting a balanced diet and assisting users to understand their meal consumption. In this paper, we propose a novel Lightweight Neural Architecture Search (LNAS) model to self-generate a thin Convolutional Neural Network (CNN) that can be executed on mobile devices with limited processing power. LNAS has a sophisticated search space and modern search strategy to design a child model with reinforcement learning. Extensive experiments have been conducted to evaluate the model generated by LNAS, namely LNAS-NET. The experimental result shows that the proposed LNAS-NET outperformed the state-of-the-art lightweight models in terms of training speed and accuracy metric. Those experiments indicate the effectiveness of LNAS without sacrificing the model performance. It provides a good direction to move toward the era of AutoML and mobile-friendly neural model design.

Download Full-text

Automatic Number Plate Recognition System using Connected Component Analysis and Convolutional Neural Network

International Journal of Engineering and Advanced Technology - Regular Issue ◽

10.35940/ijeat.f1636.1011121 ◽

2021 ◽

Vol 11 (1) ◽

pp. 167-173

Author(s):

Sneh Kanwar Singh ◽

◽

Dr. Raman Maini ◽

Dr. Dhavlessh Ratan ◽

◽

...

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Technological Progress ◽

Recognition System ◽

Experimental Result ◽

License Plate ◽

Employee Productivity ◽

Connected Component ◽

Principal Function ◽

Video Frames

Technology is becoming constantly important for customers. Automatic number plate Recognition (ANPR) is a device which enables the identification of a number plate in real time. For an intelligent car service, ANPR helps to promote growth, customize the classic app and increase consumer and employee productivity. Within the specification, the principal function of ANPR lies of removing the characteristics from an illustration of a license plate. An application that enables customers to display automobile repairs through the license platform number only derived from a loaded picture is augmented by a smart car service. Technological progress is that, so it is thought that improvement is important in this region too, so the best choice for automotive services is a smart car company. This work proposed a methodology to detect the numbers from car license plate using convolutional neural network. In the preprocessing of photographs on license plates, the WLS and FFT filters were included. The images are then fed into the convolutional trainings neural network. On more plates and tests is reported during the testing. Therefore, the findings indicate that the proposed solution can be taken in less time from the license model to accurately identify the characters. The experimental result shows the significance of proposed research by achieving an accuracy of 98% for the localization and true recognition of license plates from the video frames.

Download Full-text

A Biologically Inspired Evolving Spiking Neural Model with Rank-Order Population Coding and a Taste Recognition System Case Study

System and Circuit Design for Biologically-Inspired Intelligent Learning ◽

10.4018/978-1-60960-018-1.ch007 ◽

2011 ◽

pp. 136-155 ◽

Cited By ~ 1

Author(s):

S. Soltic ◽

N. Kasabov

Keyword(s):

Neural Network ◽

Adaptive Learning ◽

Rank Order ◽

Neural Model ◽

Recognition System ◽

Population Coding ◽

Spiking Neural Network ◽

Biologically Inspired ◽

Information Encoding ◽

Evolving Spiking Neural Network

The human brain has an amazing ability to recognize hundreds of thousands of different tastes. The question is: can we build artificial systems that can achieve this level of complexity? Such systems would be useful in biosecurity, the chemical and food industry, security, in home automation, etc. The purpose of this chapter is to explore how spiking neurons could be employed for building biologically plausible and efficient taste recognition systems. It presents an approach based on a novel spiking neural network model, the evolving spiking neural network with population coding (ESNN-PC), which is characterized by: (i) adaptive learning, (ii) knowledge discovery and (iii) accurate classification. ESNN-PC is used on a benchmark taste problem where the effectiveness of the information encoding, the quality of extracted rules and the model’s adaptive properties are explored. Finally, applications of ESNN-PC in recognition of the increasing interest in robotics and pervasive computing are suggested.

Download Full-text

Neural Network for Earthquake Prediction Based on Automatic Clustering in Indonesia

JOIV International Journal on Informatics Visualization ◽

10.30630/joiv.2.1.106 ◽

2018 ◽

Vol 2 (1) ◽

pp. 37 ◽

Cited By ~ 1

Author(s):

Mohammad Nur Shodiq ◽

Dedy Hidayat Kusuma ◽

Mirza Ghulam Rifqi ◽

Ali Ridho Barakbah ◽

Tri Harsono

Keyword(s):

Neural Network ◽

Quality Evaluation ◽

Model Performance ◽

Optimal Number ◽

Hill Climbing ◽

Experimental Result ◽

Earthquake Occurrence ◽

Hill Climbing Algorithm ◽

Automatic Clustering ◽

Proposed Model

A model of artiﬁcial neural networks (ANNs) is presented in this paper to predict aftershock during the next five days after an earthquake occurrence in selected cluster of Indonesia with magnitude equal or larger than given threshold. The data were obtained from Indonesian Agency for Meteorological, Climatological and Geophysics (BMKG) and United States Geological Survey’s (USGS). Six clusters was an optimal number of cluster base-on cluster analysis implementing Valley Tracing and Hill Climbing algorithm, while Hierarchical K-means was applied for datasets clustering. A quality evaluation was then conducted to measure the proposed model performance for two different thresholds. The experimental result shows that the model gave better performance for predicting an aftershock occurrence that equal or larger than 6 Richter’s scale magnitude.

Download Full-text

Music recognition algorithm based on T-S cognitive neural network

Translational Neuroscience ◽

10.1515/tnsci-2019-0023 ◽

2019 ◽

Vol 10 (1) ◽

pp. 135-140 ◽

Cited By ~ 1

Author(s):

Fei Yan

Keyword(s):

Neural Network ◽

Relevant Information ◽

Recognition System ◽

Recognition Algorithm ◽

Experimental Result ◽

Function Parameter ◽

Main Task ◽

Improved Genetic Algorithm ◽

Compensation Factor ◽

Music Recognition

Abstract The main task of music recognition is to acquire relevant information of music content through processing and feature extraction of audio signals, and then used for comparison, classification, and automatic recording. The cognitive neural network based on T-S model is used to train the network weights with improved genetic algorithm in the paper. The strategy of membership function parameter adjustment is combined with the combination of momentum method and learning rate adaptive adjustment. The new proposed algorithm can be used in the music recognition algorithm by adding a compensation factor related to the input dimension on the membership degree, and the experimental result of the rule disaster caused by the excessive input dimension shows that the new proposed method can be applied to the music recognition system. At the same time, it shows that the accuracy rate of the recognition network is more accurate than that of the other algorithms, and its robustness is better.

Download Full-text

Binarized Neural Architecture Search

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v34i07.6624 ◽

2020 ◽

Vol 34 (07) ◽

pp. 10526-10533 ◽

Cited By ~ 1

Author(s):

Hanlin Chen ◽

Li'an Zhuo ◽

Baochang Zhang ◽

Xiawu Zheng ◽

Jianzhuang Liu ◽

...

Keyword(s):

Neural Network ◽

Neural Networks ◽

State Of The Art ◽

Optimization Methods ◽

Search Space ◽

Network Architectures ◽

Neural Architecture ◽

Space Reduction ◽

The Cost ◽

A Performance

Neural architecture search (NAS) can have a significant impact in computer vision by automatically designing optimal neural network architectures for various tasks. A variant, binarized neural architecture search (BNAS), with a search space of binarized convolutions, can produce extremely compressed models. Unfortunately, this area remains largely unexplored. BNAS is more challenging than NAS due to the learning inefficiency caused by optimization requirements and the huge architecture space. To address these issues, we introduce channel sampling and operation space reduction into a differentiable NAS to significantly reduce the cost of searching. This is accomplished through a performance-based strategy used to abandon less potential operations. Two optimization methods for binarized neural networks are used to validate the effectiveness of our BNAS. Extensive experiments demonstrate that the proposed BNAS achieves a performance comparable to NAS on both CIFAR and ImageNet databases. An accuracy of 96.53% vs. 97.22% is achieved on the CIFAR-10 dataset, but with a significantly compressed model, and a 40% faster search than the state-of-the-art PC-DARTS.

Download Full-text

Research on sports action recognition system based on cluster regression and improved ISA deep network

Journal of Intelligent & Fuzzy Systems ◽

10.3233/jifs-189062 ◽

2020 ◽

Vol 39 (4) ◽

pp. 5871-5881

Author(s):

Hui Xu ◽

Rong Yan

Keyword(s):

Neural Network ◽

Action Recognition ◽

Recognition Rate ◽

Model Performance ◽

Recognition System ◽

Motion Recognition ◽

Deep Network ◽

Smart Technologies ◽

Human Behavior Recognition ◽

Video Library

The movements of sports athletes are complex and difficult to identify with current smart technologies. Therefore, in order to improve the sports athlete recognition rate, this paper analyze the sports action recognition system based on cluster regression and improved ISA deep network. Through literature investigation, this paper chooses ISA neural network as the basis of the algorithm. At the same time, this paper analyzes the shortcomings of traditional ISA neural network, combines the sports player’s motion recognition requirements to improve the traditional ISA neural network, and builds a sports player motion recognition system based on the improved ISA neural network algorithm. In addition, this paper uses the network data collection method to construct the sports player action video library and takes the basketball project as an example for analysis and identifies it through feature judgment. Finally, this paper builds experiments to perform model performance analysis. The research shows that the recognition rate of basketball action is greatly improved compared with the traditional algorithm model, the results verify that the improved ISA deep network proposed in this paper has significant effectiveness in the field of human behavior recognition research.

Download Full-text

Real-Time Mask Recognition

IoT ◽

10.3390/iot2040035 ◽

2021 ◽

Vol 2 (4) ◽

pp. 688-716

Author(s):

Rachel M. Billings ◽

Alan J. Michaels

Keyword(s):

Neural Network ◽

Low Power ◽

Real Time ◽

Low Cost ◽

Model Performance ◽

Network Models ◽

Recognition System ◽

Neural Network Models ◽

Still Images ◽

Experimental Dataset

While a variety of image processing studies have been performed to quantify the potential performance of neural network-based models using high-quality still images, relatively few studies seek to apply those models to a real-time operational context. This paper seeks to extend prior work in neural-network-based mask detection algorithms to a real-time, low-power deployable context that is conducive to immediate installation and use. Particularly relevant in the COVID-19 era with varying rules on mask mandates, this work applies two neural network models to inference of mask detection in both live (mobile) and recorded scenarios. Furthermore, an experimental dataset was collected where individuals were encouraged to use presentation attacks against the algorithm to quantify how perturbations negatively impact model performance. The results from evaluation on the experimental dataset are further investigated to identify the degradation caused by poor lighting and image quality, as well as to test for biases within certain demographics such as gender and ethnicity. In aggregate, this work validates the immediate feasibility of a low-power and low-cost real-time mask recognition system.

Download Full-text

A modular neural network linking Hyper RBF and AVITE models for reaching moving objects

Robotica ◽

10.1017/s0263574704001055 ◽

2005 ◽

Vol 23 (5) ◽

pp. 625-633 ◽

Cited By ~ 6

Author(s):

J. L. Pedreño-Molina ◽

J. Molina-Vilaplana ◽

J. López-Coronado ◽

P. Gorce

Keyword(s):

Neural Network ◽

Visual Information ◽

Moving Objects ◽

Neural Model ◽

Inverse Kinematic ◽

Robotic Systems ◽

Vector Integration ◽

Neural Architecture ◽

Tracking Tasks ◽

The Difference

In this paper, the problem of precision reaching applications in robotic systems for scenarios with static and non-static objects has been considered and a solution based on a modular neural architecture has been proposed and implemented. The goal of this solution is to combine robustness and capability mapping trajectories from two biologically plausible neural network sub-modules: Hyper RBF and AVITE. The Hyper Basis Radial Function (HypRBF) neural network solves the inverse kinematic in redundant robotic systems, while the Adaptive Vector Integration to End-Point (AVITE) visuo-motor neural model quickly maps the difference vector between current and desired position in both spatial (visual information) and motor coordinates (propioceptive information). The anthropomorphic behaviour of the proposed architecture for reaching and tracking tasks in presence of spatial perturbations has been validated over a real arm-head robotic platform.

Download Full-text