Search results for: neural network classification

Authors: Himanshu Payal, Sachin Maheshwari, Pushpendra S. Bharti

Abstract:

Electrical discharge machining (EDM), a non-conventional machining process, finds wide applications for shaping difficult-to-cut alloys. Process modeling of EDM is required to exploit the process to the fullest. Process modeling of EDM is a challenging task owing to involvement of so many electrical and non-electrical parameters. This work is an attempt to model the EDM process using artificial neural network (ANN). Experiments were carried out on die-sinking EDM taking Inconel 825 as work material. ANN modeling has been performed using experimental data. The prediction ability of trained network has been verified experimentally. Results indicate that ANN can predict the values of performance measures of EDM satisfactorily.

Keywords: artificial neural network, EDM, metal removal rate, modeling, surface roughness

Procedia PDF Downloads 382

Authors: Emad Amar, Tawfik Khattab, Fatma Zada

Abstract:

Predicting earthquakes is an important issue in the study of geography. Accurate prediction of earthquakes can help people to take effective measures to minimize the loss of personal and economic damage, such as large casualties, destruction of buildings and broken of traffic, occurred within a few seconds. United States Geological Survey (USGS) science organization provides reliable scientific information of Earthquake Existed throughout history & Preliminary database from the National Center Earthquake Information (NEIC) show some useful factors to predict an earthquake in a seismic area like Aleutian Arc in the U.S. state of Alaska. The main advantage of this prediction method that it does not require any assumption, it makes prediction according to the future evolution of object's time series. The article compares between simulation data result from trained BP and RBF neural network versus actual output result from the system calculations. Therefore, this article focuses on analysis of data relating to real earthquakes. Evaluation results show better accuracy and higher speed by using radial basis functions (RBF) neural network.

Keywords: BP neural network, prediction, RBF neural network, earthquake

Procedia PDF Downloads 464

Authors: Lu Zhao, Nadir Farhi, Yeltsin Valero, Zoi Christoforou, Nadia Haddadou

Abstract:

In this paper, a long short-term memory (LSTM) neural network model is proposed to replicate simultaneously car-following and lane-changing behaviors in road networks. By combining two kinds of LSTM layers and three input designs of the neural network, six variants of the LSTM model have been created. These models were trained and tested on the NGSIM 101 dataset, and the results were evaluated in terms of longitudinal speed and lateral position, respectively. Then, we compared the LSTM model with a classical car-following model (the intelligent driving model (IDM)) in the part of speed decision. In addition, the LSTM model is compared with a model using classical neural networks. After the comparison, the LSTM model demonstrates higher accuracy than the physical model IDM in terms of car-following behavior and displays better performance with regard to both car-following and lane-changing behavior compared to the classical neural network model.

Keywords: traffic modeling, neural networks, LSTM, car-following, lane-change

Procedia PDF Downloads 218

Authors: Frankie Burgos, Emely Munar, Conrado Basa

Abstract:

This paper aimed at developing an artificial neural network (ANN) model specifically for the Philippine Stock Exchange index closing value. The inputs to the ANN are US Dollar and Philippine Peso(USD-PHP) exchange rate, GDP growth of the country, quarterly inflation rate, 10-year bond yield, credit rating of the country, previous open, high, low, close values and volume of trade of the Philippine Stock Exchange Index (PSEi), gold price of the previous day, National Association of Securities Dealers Automated Quotations (NASDAQ), Standard and Poor’s 500 (S & P 500) and the iShares MSCI Philippines ETF (EPHE) previous closing value. The target is composed of the closing value of the PSEi during the 627 trading days from November 3, 2011, to May 30, 2014. MATLAB’s Neural Network toolbox was employed to create, train and simulate the network using multi-layer feed forward neural network with back-propagation algorithm. The results satisfactorily show that the neural network developed has the ability to model the PSEi, which is affected by both internal and external economic factors. It was found out that the inputs used are the main factors that influence the movement of the PSEi closing value.

Keywords: artificial neural networks, artificial intelligence, philippine stocks exchange index, stocks trading

Procedia PDF Downloads 265

Authors: Chia-Hao Chang, Geng-Gui Wang, Jee-Cheng Wu

Abstract:

The prediction of a landslide is a difficult task because it requires a detailed study of past activities using a complete range of investigative methods to determine the changing condition. In this research, first step, LiDAR 1-meter by 1-meter resolution of digital elevation model (DEM) was used to generate six environmental factors of landslide. Then, back-propagation neural networks (BPNN) was adopted to identify scarp, landslide areas and non-landslide areas. The BPNN uses 6 environmental factors in input layer and 1 output layer. Moreover, 6 landslide areas are used as training areas and 4 landslide areas as test areas in the BPNN. The hidden layer is set to be 1 and 2; the hidden layer neurons are set to be 4, 5, 6, 7 and 8; the learning rates are set to be 0.01, 0.1 and 0.5. When using 1 hidden layer with 7 neurons and the learning rate sets to be 0.5, the result of Network training root mean square error is 0.001388. Finally, evaluation of BPNN classification accuracy by the confusion matrix shows that the overall accuracy can reach 94.4%, and the Kappa value is 0.7464.

Keywords: digital elevation model, DEM, environmental factors, back-propagation neural network, BPNN, LiDAR

Procedia PDF Downloads 110

Authors: Gulfam Haider, sana danish

Abstract:

Convolutional Neural Networks (CNNs) have emerged as powerful tools for image classification, and the choice of optimizers profoundly affects their performance. The study of optimizers and their adaptations remains a topic of significant importance in machine learning research. While numerous studies have explored and advocated for various optimizers, the efficacy of these optimization techniques is still subject to scrutiny. This work aims to address the challenges surrounding the effectiveness of optimizers by conducting a comprehensive analysis and evaluation. The primary focus of this investigation lies in examining the performance of different optimizers when employed in conjunction with the popular activation function, Rectified Linear Unit (ReLU). By incorporating ReLU, known for its favorable properties in prior research, the aim is to bolster the effectiveness of the optimizers under scrutiny. Specifically, we evaluate the adjustment of these optimizers with both the original Softmax activation function and the modified ReLU activation function, carefully assessing their impact on overall performance. To achieve this, a series of experiments are conducted using a well-established benchmark dataset for image classification tasks, namely the Canadian Institute for Advanced Research dataset (CIFAR-10). The selected optimizers for investigation encompass a range of prominent algorithms, including Adam, Root Mean Squared Propagation (RMSprop), Adaptive Learning Rate Method (Adadelta), Adaptive Gradient Algorithm (Adagrad), and Stochastic Gradient Descent (SGD). The performance analysis encompasses a comprehensive evaluation of the classification accuracy, convergence speed, and robustness of the CNN models trained with each optimizer. Through rigorous experimentation and meticulous assessment, we discern the strengths and weaknesses of the different optimization techniques, providing valuable insights into their suitability for image classification tasks. By conducting this in-depth study, we contribute to the existing body of knowledge surrounding optimizers in CNNs, shedding light on their performance characteristics for image classification. The findings gleaned from this research serve to guide researchers and practitioners in making informed decisions when selecting optimizers and activation functions, thus advancing the state-of-the-art in the field of image classification with convolutional neural networks.

Keywords: deep neural network, optimizers, RMsprop, ReLU, stochastic gradient descent

Procedia PDF Downloads 62

Authors: Firas M. Tuaimah

Abstract:

This paper presents an Artificial Neural Network based approach for short-term load forecasting and exactly for two days ahead. Two seasons have been discussed for Iraqi power system, namely summer and winter; the hourly load demand is the most important input variables for ANN based load forecasting. The recorded daily load profile with a lead time of 1-48 hours for July and December of the year 2012 was obtained from the operation and control center that belongs to the Ministry of Iraqi electricity. The results of the comparison show that the neural network gives a good prediction for the load forecasting and for two days ahead.

Keywords: short-term load forecasting, artificial neural networks, back propagation learning, hourly load demand

Procedia PDF Downloads 428

Authors: Trupti Chavan, Ramachandra Guda, Kameshwar Rao

Abstract:

The lack of data is a pain problem in medical image analysis using a convolutional neural network (CNN). This work uses various spatial transformation techniques to address the medical image augmentation issue for knee detection and localization using an enhanced single shot detector (SSD) network. The spatial transforms like a negative, histogram equalization, power law, sharpening, averaging, gaussian blurring, etc. help to generate more samples, serve as pre-processing methods, and highlight the features of interest. The experimentation is done on the OpenKnee dataset which is a collection of knee images from the openly available online sources. The CNN called enhanced single shot detector (SSD) is utilized for the detection and localization of the knee joint from a given X-ray image. It is an enhanced version of the famous SSD network and is modified in such a way that it will reduce the number of prediction boxes at the output side. It consists of a classification network (VGGNET) and an auxiliary detection network. The performance is measured in mean average precision (mAP), and 99.96% mAP is achieved using the proposed enhanced SSD with spatial transformations. It is also seen that the localization boundary is comparatively more refined and closer to the ground truth in spatial augmentation and gives better detection and localization of knee joints.

Keywords: data augmentation, enhanced SSD, knee detection and localization, medical image analysis, openKnee, Spatial transformations

Procedia PDF Downloads 129

Authors: K. D. Sharma, M. Parida, S. S. Jain, Anju Saini, V. K. Katiyar

Abstract:

Air quality models form one of the most important components of an urban air quality management plan. Various statistical modeling techniques (regression, multiple regression and time series analysis) have been used to predict air pollution concentrations in the urban environment. These models calculate pollution concentrations due to observed traffic, meteorological and pollution data after an appropriate relationship has been obtained empirically between these parameters. Artificial neural network (ANN) is increasingly used as an alternative tool for modeling the pollutants from vehicular traffic particularly in urban areas. In the present paper, an attempt has been made to model traffic air pollution, specifically CO concentration using neural networks. In case of CO concentration, two scenarios were considered. First, with only classified traffic volume input and the second with both classified traffic volume and meteorological variables. The results showed that CO concentration can be predicted with good accuracy using artificial neural network (ANN).

Keywords: air quality management, artificial neural network, meteorological variables, statistical modeling

Procedia PDF Downloads 491

Authors: Debneil Saha Roy

Abstract:

Weather forecasting is a complex topic and is well suited for analysis by deep learning approaches. With the wide availability of weather observation data nowadays, these approaches can be utilized to identify immediate comparisons between historical weather forecasts and current observations. This work explores the application of deep learning techniques to weather forecasting in order to accurately predict the weather over a given forecast hori­zon. Three deep neural networks are used in this study, namely, Multi-Layer Perceptron (MLP), Long Short Tunn Memory Network (LSTM) and a combination of Convolutional Neural Network (CNN) and LSTM. The predictive performance of these models is compared using two evaluation metrics. The results show that forecasting accuracy increases with an increase in the complexity of deep neural networks.

Keywords: convolutional neural network, deep learning, long short term memory, multi-layer perceptron

Procedia PDF Downloads 137

Authors: Venkatesh Pulletikurthi, Karthik B. Ariyur, Luciano Castillo

Abstract:

The system identification from the turbulence wakes will lead to the tactical advantage to prepare and also, to predict the trajectory of the opponents’ movements. A low-order modeling technique, POD, is used to predict the object based on the wake pattern and compared with pre-trained image recognition neural network (NN) to classify the wake patterns into objects. It is demonstrated that low-order modeling, POD, is able to predict the objects better compared to pretrained NN by ~30%.

Keywords: the bluff body wakes, low-order modeling, neural network, system identification

Procedia PDF Downloads 146

Authors: Hrishi Rakshit, Pooneh Bagheri Zadeh

Abstract:

This paper investigates the performance of drone classification methods using conventional DCNN with different hyperparameters, when additional drone audio data is embedded in the dataset for training and further classification. In this paper, first a custom dataset is created using different images of drones from University of South California (USC) datasets and Leeds Beckett university datasets with embedded drone audio signal. The three well-known DCNN architectures namely, Resnet50, Darknet53 and Shufflenet are employed over the created dataset tuning their hyperparameters such as, learning rates, maximum epochs, Mini Batch size with different optimizers. Precision-Recall curves and F1 Scores-Threshold curves are used to evaluate the performance of the named classification algorithms. Experimental results show that Resnet50 has the highest efficiency compared to other DCNN methods.

Keywords: drone classifications, deep convolutional neural network, hyperparameters, drone audio signal

Procedia PDF Downloads 56

Authors: Debajyoti Mukherjee, Arathy B. S., Arpita Sahu, Saranga P. Pogula

Abstract:

A Functional Model to mimic the functional correctness of a Neural Network Compute Accelerator IP is very crucial for design validation. Neural network workloads are based on a Brain Floating Point (BF-16) data type. The major challenge we were facing was the incompatibility of gcc compilers to BF-16 datatype, which we addressed with a native BF-16 generator integrated to our functional model. Moreover, working with big GEMM (General Matrix Multiplication) or SpMM (Sparse Matrix Multiplication) Work Loads (Dense or Sparse) and debugging the failures related to data integrity is highly painstaking. In this paper, we are addressing the quality challenge of such a complex Neural Network Accelerator design by proposing a Functional Model-based scoreboard or Software model using SystemC. The proposed Functional Model executes the assembly code based on the ISA of the processor IP, decodes all instructions, and executes as expected to be done by the DUT. The said model would give a lot of visibility and debug capability in the DUT bringing up micro-steps of execution.

Keywords: ISA (instruction set architecture), NN (neural network), TLM (transaction-level modeling), GEMM (general matrix multiplication)

Procedia PDF Downloads 48

Authors: Tharini N. de Silva, Xiao Zhibo, Zhao Rui, Mao Kezhi

Abstract:

Causal relation identification is a crucial task in information extraction and knowledge discovery. In this work, we present two approaches to causal relation identification. The first is a classification model trained on a set of knowledge-based features. The second is a deep learning based approach training a model using convolutional neural networks to classify causal relations. We experiment with several different convolutional neural networks (CNN) models based on previous work on relation extraction as well as our own research. Our models are able to identify both explicit and implicit causal relations as well as the direction of the causal relation. The results of our experiments show a higher accuracy than previously achieved for causal relation identification tasks.

Keywords: causal realtion extraction, relation extracton, convolutional neural network, text representation

Procedia PDF Downloads 685

Authors: Si Mon Kueh, Tom J. Kazmierski

Abstract:

There are about 1% of the world population suffering from the hidden disability known as epilepsy and major developing countries are not fully equipped to counter this problem. In order to reduce the inconvenience and danger of epilepsy, different methods have been researched by using a artificial neural network (ANN) classification to distinguish epileptic waveforms from normal brain waveforms. This paper outlines the aim of achieving massive ANN parallelization through a dedicated hardware using bit-serial processing. The design of this bit-serial Neural Processing Element (NPE) is presented which implements the functionality of a complete neuron using variable accuracy. The proposed design has been tested taking into consideration non-idealities of a hardware ANN. The NPE consists of a bit-serial multiplier which uses only 16 logic elements on an Altera Cyclone IV FPGA and a bit-serial ALU as well as a look-up table. Arrays of NPEs can be driven by a single controller which executes the neural processing algorithm. In conclusion, the proposed compact NPE design allows the construction of complex hardware ANNs that can be implemented in a portable equipment that suits the needs of a single epileptic patient in his or her daily activities to predict the occurrences of impending tonic conic seizures.

Keywords: Artificial Neural Networks (ANN), bit-serial neural processor, FPGA, Neural Processing Element (NPE)

Procedia PDF Downloads 292

Authors: Alireza Lavaei, Alireza Lohrasbi, Mohammadali M. Shahlaei

Abstract:

To reduce the overall time of structural optimization for earthquake loads two strategies are adopted. In the first strategy, a neural system consisting self-organizing map and radial basis function neural networks, is utilized to predict the time history responses. In this case, the input space is classified by employing a self-organizing map neural network. Then a distinct RBF neural network is trained in each class. In the second strategy, an improved genetic algorithm is employed to find the optimum design. A 72-bar space truss is designed for optimal weight using exact and approximate analysis for the El Centro (S-E 1940) earthquake loading. The numerical results demonstrate the computational advantages and effectiveness of the proposed method.

Keywords: optimization, genetic algorithm, neural networks, self-organizing map

Procedia PDF Downloads 273

Authors: Kishore K. Pochampally

Abstract:

The success of a supply chain heavily relies on the efficiency of the suppliers involved. In this paper, we propose a neural network approach to evaluate the efficiency of a supplier, which is being considered for inclusion in a supply chain, using the available linguistic (fuzzy) data of suppliers that already exist in the supply chain. The approach is carried out in three phases, as follows: In phase one, we identify criteria for evaluation of the supplier of interest. Then, in phase two, we use performance measures of already existing suppliers to construct a neural network that gives weights (importance values) of criteria identified in phase one. Finally, in phase three, we calculate the overall rating of the supplier of interest. The following are the major findings of the research conducted for this paper: (i) linguistic (fuzzy) ratings of suppliers such as 'good', 'bad', etc., can be converted (defuzzified) to numerical ratings (1 – 10 scale) using fuzzy logic so that those ratings can be used for further quantitative analysis; (ii) it is possible to construct and train a multi-level neural network in order to determine the weights of the criteria that are used to evaluate a supplier; and (iii) Borda’s rule can be used to group the weighted ratings and calculate the overall efficiency of the supplier.

Keywords: fuzzy data, neural network, supplier, supply chain

Procedia PDF Downloads 85

Authors: Daham Owaid Matrood, Naqaa Hussein Raheem

Abstract:

Evolutionary algorithms and Artificial neural networks (ANN) are two relatively young research areas that were subject to a steadily growing interest during the past years. This paper examines the use of Particle Swarm Optimization (PSO) to train a multi-layer feed forward neural network for demand forecasting. We use in this paper weekly demand data for packed cement and towels, which have been outfitted by the Northern General Company for Cement and General Company of prepared clothes respectively. The results showed superiority of trained neural networks using particle swarm optimization on neural networks trained using error back propagation because their ability to escape from local optima.

Keywords: artificial neural network, demand forecasting, particle swarm optimization, weight optimization

Procedia PDF Downloads 411

Authors: John Gibbs, Benjamin Flanders, Dylan Pozorski, Weixuan Liu

Abstract:

Weight Painting–adjusting the influence a skeletal joint has on a given vertex in a character mesh–is an arduous and time con- suming part of the 3D animation pipeline. This process generally requires a trained technical animator and many hours of work to complete. Our skiNNer plug-in, which works within Autodesk’s Maya 3D animation software, uses Machine Learning and data pro- cessing techniques to create a deep neural network model that can accomplish the weight painting task in seconds rather than hours for bipedal quasi-humanoid character meshes. In order to create a properly trained network, a number of challenges were overcome, including curating an appropriately large data library, managing an arbitrary 3D mesh size, handling arbitrary skeletal architectures, accounting for extreme numeric values (most data points are near 0 or 1 for weight maps), and constructing an appropriate neural network model that can properly capture the high frequency alter- ation between high weight values (near 1.0) and low weight values (near 0.0). The arrived at neural network model is a cross between a traditional CNN, deep residual network, and fully dense network. The resultant network captures the unusually hard-edged features of a weight map matrix, and produces excellent results on many bipedal models.

Keywords: 3d animation, animation, character, rigging, skinning, weight painting, machine learning, artificial intelligence, neural network, deep neural network

Procedia PDF Downloads 240

Authors: Zhaofeng Li, Jun Kang Chow, Yu-Hsing Wang

Abstract:

This paper attempts to bridge the soil properties and the mechanical response of soil in the discrete element method (DEM) simulation. The artificial neural network (ANN) was therefore adopted, aiming to reproduce the stress-strain-volumetric response when soil properties are given. 31 biaxial shearing tests with varying soil parameters (e.g., initial void ratio and interparticle friction coefficient) were generated using the DEM simulations. Based on these 45 sets of training data, a three-layer neural network was established which can output the entire stress-strain-volumetric curve during the shearing process from the input soil parameters. Beyond the training data, 2 additional sets of data were generated to examine the validity of the network, and the stress-strain-volumetric curves for both cases were well reproduced using this network. Overall, the ANN was found promising in predicting the soil behavior and reducing repetitive simulation work.

Keywords: artificial neural network, discrete element method, soil properties, stress-strain-volumetric response

Procedia PDF Downloads 369

Authors: Nasrin Bakhshizadeh, Ashkan Forootan

Abstract:

A major problem that affects the quality control of polymer in the industrial polymerization is the lack of suitable on-line measurement tools to evaluate the properties of the polymer such as melt and density indices. Controlling the polymerization in ordinary method is performed manually by taking samples, measuring the quality of polymer in the lab and registry of results. This method is highly time consuming and leads to producing large number of incompatible products. An online application for estimating melt index and density proposed in this study is a neural network based on the input-output data of the polyethylene production plant. Temperature, the level of reactors' bed, the intensity of ethylene mass flow, hydrogen and butene-1, the molar concentration of ethylene, hydrogen and butene-1 are used for the process to establish the neural model. The neural network is taught based on the actual operational data and back-propagation and Levenberg-Marquart techniques. The simulated results indicate that the neural network process model established with three layers (one hidden layer) for forecasting the density and the four layers for the melt index is able to successfully predict those quality properties.

Keywords: polyethylene, polymerization, density, melt index, neural network

Procedia PDF Downloads 119

Authors: Olivia Borgue, Massimo Panarotto, Ola Isaksson

Abstract:

This article presents a neural networks-based strategy for reducing the need for support structures when designing for additive manufacturing (AM). Additive manufacturing is a relatively new and immature industrial technology, and the information to make confident decisions when designing for AM is limited. This lack of information impacts especially the early stages of engineering design, for instance, it is difficult to actively consider the support structures needed for manufacturing a part. This difficulty is related to the challenge of designing a product geometry accounting for customer requirements, manufacturing constraints and minimization of support structure. The approach presented in this article proposes an automatized geometry modification technique for reducing the use of the support structures while designing for AM. This strategy starts with a neural network-based strategy for shape recognition to achieve product classification, using an STL file of the product as input. Based on the classification, an automatic part geometry modification based on MATLAB© is implemented. At the end of the process, the strategy presents different geometry modification alternatives depending on the type of product to be designed. The geometry alternatives are then evaluated adopting a QFD-like decision support tool.

Keywords: additive manufacturing, engineering design, geometry modification optimization, neural networks

Procedia PDF Downloads 222

Authors: Rakkrit Duangsoithong, Jermphiphut Jaruenpunyasak, Alba Garcia

Abstract:

The use of foot recognition can be applied in many medical fields such as the gait pattern analysis and the knee exercises of patients in rehabilitation. Generally, a camera-based foot recognition system is intended to capture a patient image in a controlled room and background to recognize the foot in the limited views. However, this system can be inconvenient to monitor the knee exercises at home. In order to overcome these problems, this paper proposes to use the deep learning method using Convolutional Neural Networks (CNNs) for foot recognition. The results are compared with the traditional classification method using LBP and HOG features with kNN and SVM classifiers. According to the results, deep learning method provides better accuracy but with higher complexity to recognize the foot images from online databases than the traditional classification method.

Keywords: foot recognition, deep learning, knee rehabilitation, convolutional neural network

Procedia PDF Downloads 126

Authors: Pavan K. Rallabandi, Kailash C. Patidar

Abstract:

In this paper, we present an optimization technique or a learning algorithm using the hybrid architecture by combining the most popular sequence recognition models such as Recurrent Neural Networks (RNNs) and Hidden Markov models (HMMs). In order to improve the sequence or pattern recognition/ classification performance by applying a hybrid/neural symbolic approach, a gradient descent learning algorithm is developed using the Real Time Recurrent Learning of Recurrent Neural Network for processing the knowledge represented in trained Hidden Markov Models. The developed hybrid algorithm is implemented on automata theory as a sample test beds and the performance of the designed algorithm is demonstrated and evaluated on learning the deterministic finite state automata.

Keywords: hybrid systems, hidden markov models, recurrent neural networks, deterministic finite state automata

Procedia PDF Downloads 355

Authors: Christopher Lama, Alix Rieser, Aleksandra Molchanova, Charles Thangaraj

Abstract:

New technologies like Tesla’s Dojo have made high-performance embedded computing more available. Although automobile computing has developed and benefited enormously from these more recent technologies, the costs are still high, prohibitively high in some cases for broader adaptation, particularly for the after-market and enthusiast markets. This project aims to implement a Raspberry Pi-based low-power (under one hundred Watts) highly mobile computing cluster for a neural network. The computing cluster built from off-the-shelf components is more affordable and, therefore, makes wider adoption possible. The paper describes the design of the neural network, Raspberry Pi-based cluster, and applications the cluster will run. The neural network will use input data from sensors and cameras to project a live view of the road state as the user drives. The neural network will be trained to predict traffic behavior and generate warnings when potentially dangerous situations are predicted. The significant outcomes of this study will be two folds, firstly, to implement and test the low-cost cluster, and secondly, to ascertain the effectiveness of the predictive AI implemented on the cluster.

Keywords: CS pedagogy, student research, cluster computing, machine learning

Procedia PDF Downloads 57

Authors: Rujia Chen, Ajit Narayanan

Abstract:

Convolutional neural networks (CNN), as typically applied in deep learning, use layer-wise backpropagation (BP) to construct filters and kernels for feature extraction. Such filters are 2D or 3D groups of weights for constructing feature maps at subsequent layers of the CNN and are shared across the entire input. BP as a gradient descent algorithm has well-known problems of getting stuck at local optima. The use of genetic algorithms (GAs) for evolving weights between layers of standard artificial neural networks (ANNs) is a well-established area of neuroevolution. In particular, the use of crossover techniques when optimizing weights can help to overcome problems of local optima. However, the application of GAs for evolving the weights of filters and kernels in CNNs is not yet an established area of neuroevolution. In this paper, a GA-based filter development algorithm is proposed. The results of the proof-of-concept experiments described in this paper show the proposed GA algorithm can find filter weights through evolutionary techniques rather than BP learning. For some simple classification tasks like geometric shape recognition, the proposed algorithm can achieve 100% accuracy. The results for MNIST classification, while not as good as possible through standard filter learning through BP, show that filter and kernel evolution warrants further investigation as a new subarea of neuroevolution for deep architectures.

Keywords: neuroevolution, convolutional neural network, genetic algorithm, filters, kernels

Procedia PDF Downloads 157

Authors: Sule Yucelbas, Gulay Tezel, Cuneyt Yucelbas, Seral Ozsen

Abstract:

In this study, it was tried to identify some heart rhythm disorders by electrocardiography (ECG) data that is taken from MIT-BIH arrhythmia database by subtracting the required features, presenting to artificial neural networks (ANN), artificial immune systems (AIS), artificial neural network based on artificial immune system (AIS-ANN) and particle swarm optimization based artificial neural network (PSO-NN) classifier systems. The main purpose of this study is to evaluate the performance of hybrid AIS-ANN and PSO-ANN classifiers with regard to the ANN and AIS. For this purpose, the normal sinus rhythm (NSR), atrial premature contraction (APC), sinus arrhythmia (SA), ventricular trigeminy (VTI), ventricular tachycardia (VTK) and atrial fibrillation (AF) data for each of the RR intervals were found. Then these data in the form of pairs (NSR-APC, NSR-SA, NSR-VTI, NSR-VTK and NSR-AF) is created by combining discrete wavelet transform which is applied to each of these two groups of data and two different data sets with 9 and 27 features were obtained from each of them after data reduction. Afterwards, the data randomly was firstly mixed within themselves, and then 4-fold cross validation method was applied to create the training and testing data. The training and testing accuracy rates and training time are compared with each other. As a result, performances of the hybrid classification systems, AIS-ANN and PSO-ANN were seen to be close to the performance of the ANN system. Also, the results of the hybrid systems were much better than AIS, too. However, ANN had much shorter period of training time than other systems. In terms of training times, ANN was followed by PSO-ANN, AIS-ANN and AIS systems respectively. Also, the features that extracted from the data affected the classification results significantly.

Keywords: AIS, ANN, ECG, hybrid classifiers, PSO

Procedia PDF Downloads 410

Authors: Tianle Yang

Abstract:

As a common mental disorder, depression is a leading cause of various diseases worldwide. Early detection and treatment of depression can dramatically promote remission and prevent relapse. However, conventional ways of depression diagnosis require considerable human effort and cause economic burden, while still being prone to misdiagnosis. On the other hand, recent studies report that physical characteristics are major contributors to the diagnosis of depression, which inspires us to mine the internal relationship by neural networks instead of relying on clinical experiences. In this paper, neural networks are constructed to predict depression from physical characteristics. Two initialization methods are examined - Xaiver and Kaiming initialization. Experimental results show that a 3-layers neural network with Kaiming initialization achieves 83% accuracy.

Keywords: depression, neural network, Xavier initialization, Kaiming initialization

Procedia PDF Downloads 95

Authors: Zachary Pan

Abstract:

Our general mission is to solve the problem of classifying images into different body part types and deciding if each of them is healthy or not. However, for now, we will determine healthiness for only one-sixth of the body parts, specifically the chest. We will detect pneumonia in X-ray scans of those chest images. With this type of AI, doctors can use it as a second opinion when they are taking CT or X-ray scans of their patients. Another ad-vantage of using this machine learning classifier is that it has no human weaknesses like fatigue. The overall ap-proach to this problem is to split the problem into two parts: first, classify the image, then determine if it is healthy. In order to classify the image into a specific body part class, the body parts dataset must be split into test and training sets. We can then use many models, like neural networks or logistic regression models, and fit them using the training set. Now, using the test set, we can obtain a realistic accuracy the models will have on images in the real world since these testing images have never been seen by the models before. In order to increase this testing accuracy, we can also apply many complex algorithms to the models, like multiplicative weight update. For the second part of the problem, to determine if the body part is healthy, we can have another dataset consisting of healthy and non-healthy images of the specific body part and once again split that into the test and training sets. We then use another neural network to train on those training set images and use the testing set to figure out its accuracy. We will do this process only for the chest images. A major conclusion reached is that convolutional neural networks are the most reliable and accurate at image classification. In classifying the images, the logistic regression model, the neural network, neural networks with multiplicative weight update, neural networks with the black box algorithm, and the convolutional neural network achieved 96.83 percent accuracy, 97.33 percent accuracy, 97.83 percent accuracy, 96.67 percent accuracy, and 98.83 percent accuracy, respectively. On the other hand, the overall accuracy of the model that de-termines if the images are healthy or not is around 78.37 percent accuracy.

Keywords: body part, healthcare, machine learning, neural networks

Procedia PDF Downloads 71

Authors: Krishna Mohan Bathula, Fatou Bintou Loucoubar, FNU Kaleemunnisa, Christelle Scharff, Mark Anthony De Castro

Abstract:

Voice recognition algorithms such as automatic speech recognition and text-to-speech systems with African languages can play an important role in bridging the digital divide of Artificial Intelligence in Africa, contributing to the establishment of a fully inclusive information society. This paper proposes a Deep Learning model that can classify the user responses as inputs for an interactive voice response system. A dataset with Wolof language words ‘yes’ and ‘no’ is collected as audio recordings. A two stage Data Augmentation approach is adopted for enhancing the dataset size required by the deep neural network. Data preprocessing and feature engineering with Mel-Frequency Cepstral Coefficients are implemented. Convolutional Neural Networks (CNNs) have proven to be very powerful in image classification and are promising for audio processing when sounds are transformed into spectra. For performing voice response classification, the recordings are transformed into sound frequency feature spectra and then applied image classification methodology using a deep CNN model. The inference model of this trained and reusable Wolof voice response recognition system can be integrated with many applications associated with both web and mobile platforms.

Keywords: automatic speech recognition, interactive voice response, voice response recognition, wolof word classification

Procedia PDF Downloads 86