China
Africa
Explore chapters and articles related to this topic
Technological Development
Published in Edward Y. Uechi, Business Automation and Its Effect on the Labor Force, 2023
Technology aside, new methods and procedures have had to be developed to analyze data types that are not quantifiable. Probabilistic method is a technique that draws on probability theory. A computer that uses the probabilistic method evaluates a random variable against a set of defined conditions and gives a probability value to the random variable based on how well it matches the conditions. Another method draws on how neurons operate in the human brain. An artificial neural network (also known as a neural net) comprises nodes distinguished by input nodes, hidden nodes, and output nodes. Each node would hold a specific datum. A computer processes the nodes to find connections and the strengths of those connections among the various nodes. Machine learning, which can be divided into three areas (supervised learning, unsupervised learning, and reinforcement learning), provides a method for a computer to analyze different types of data to find a specific match or a pattern. Machine learning requires a massive amount of data, which the computer can base its conclusions on.
Proficient Prediction of Acute Lymphoblastic Leukemia Using Machine Learning Algorithm
Published in K. Gayathri Devi, Mamata Rath, Nguyen Thi Dieu Linh, Artificial Intelligence Trends for Data Analytics Using Machine Learning and Deep Learning Approaches, 2020
M. Sangeetha, K.N. Apinaya Prethi, S. Nithya
An artificial neural network is a system that is similar to the human brain in terms of structure and information processing. It is a simple model of the brain that deals with linear and nonlinear relationships between the input and output. A neural network is a collection of neurons, which is a mathematical function that gathers and processes data per the system design. The neural network can be used for any number of tasks, but it became popular for classification because of its precision. For example, face recognition can be used by our mobile devices to securing a phone from intruders. A neural network is trained with a set of images and during validation it finds out how closely a captured image matches stored images (trained data). Back-propagation plays a major role in neural network, by which high accuracy is achieved. In the back-propagation process, feedback is given to a neural network and the achieved output is compared with the output which was meant to be produced and the weighting of the neurons adjusted accordingly. This technique helps a neural network to achieve high accuracy in a shorter period than any other traditional system [1]. This makes neural networks ideal for speech recognition, text classification, language processing and semantic analysis.
Learning Based Classifiers
Published in Anil Kumar, Priyadarshi Upadhyay, A. Senthil Kumar, Fuzzy Machine Learning Algorithms for Remote Sensing Image Classification, 2020
Anil Kumar, Priyadarshi Upadhyay, A. Senthil Kumar
Artificial neural networks require large training data sets, otherwise their accuracy drops. They have been applied in various applications. Paola and Schowengerdt (1995a, 1995b) delivered an inclusive assessment of the application of the multi-layer perception in remote sensing. In today’s scenario, the data can be from disparate sources, due to which data cannot be linear, stationary, or oriented and cannot be modeled comfortably. Neural networks do not require any information about the problem type or statistical distribution of the data set. Initially developed conventional classifiers require that input data should follow a standard distribution pattern, but for neural network, this is not necessary. Learning through which parameters of algorithm is calculated is a first step in a supervised classifier. In the case of learning algorithms, learning scheme such as adjustment of weights has an altogether different meaning. In the case of networks, learning takes place with back-propagation algorithms, etc. After the network has been trained, the next step is to identify the unknown vector (image pixels) belong to which class. Training can be done in different ways, through different learning examples, learning instructions, and learning procedures. A static network is a system which follows clear-cut knowledge and manufacture phases. Networks which learn continuously during processing of data are known as energetic systems.
Shape Optimization and Flow Analysis of Supersonic Nozzles Using Deep Learning
Published in International Journal of Computational Fluid Dynamics, 2022
Aref Zanjani, Amir Mahdi Tahsini, Kimia Sadafi, Fatemeh Ghavidel Mangodeh
In this paper, artificial neural networks (ANN) and convolutional neural networks (CNN) are employed for obtaining optimum supersonic nozzle shape and flow analysis inside them respectively. Artificial neural networks (ANN) are modelled after biological neural networks and consist of interconnected nodes similar to neurons. The three key components of a neural network are the node characteristics, network topology and learning rules. Factors such as the number of inputs and outputs, weights associated with each input and output, and the activation function influence signal processing within nodes. The network's topology determines how nodes are connected, and the weights are adjusted based on learning rules. Nonlinear transfer functions are generally more effective than linear ones, as most problems are not linearly separable (Zou, Han, and So 2008). An artificial neural network is essentially a group of these nodes (neurons) gathered together in different layers with different connections. As the number of layers increases, the model enters the realm of deep artificial neural network (DANN) which can handle highly non-linear problems for the cost of increasing the overfitting risk. As will be seen later in this paper, ANN is used to predict the optimum geometry of nozzle according to desired optimisation objective. However, experience shows that stacking neurons in a fully connected layers doesn't tend to work well with image data, since the number of inputs is extremely high. Instead, using convolutional neural networks would result in a much simpler, and more accurate model.
A survey of deep learning approaches for WiFi-based indoor positioning
Published in Journal of Information and Telecommunication, 2022
Xu Feng, Khuong An Nguyen, Zhiyuan Luo
The basic structure of the systems reviewed in this section are illustrated in Figure 9. Input data, including WiFi RSS, CSI and hybrid signals from other sensors, are preprocessed before being fed into the feature extraction methods. For the best performance of the deep learning feature extraction methods, the input data are to be normalized, calibrated, augmented, classified or preprocessed with dimensionality reduction using statistical methods or traditional machine learning algorithms like SVM and PCA during preprocessing. Deep learning neural networks are used to extract hierarchical features of the input data. And the types of neural networks included in this section could be classified as Artificial Neural Networks (ANN), Convolution Neural Network (CNN), Auto Encoder (AE), Deep Belief Network (DBN), Recurrent Neural Network (RNN) and other architectures (e.g. Deep Gaussian Process (DGP)).
Study on a combined prediction method based on BP neural network and improved Verhulst model
Published in Systems Science & Control Engineering, 2019
Tong Niu, Lin Zhang, Shengjun Wei, Baoshan Zhang, Bo Zhang
The artificial neural network is a kind of artificial network composed by a large amount of simple information-processing elements called neurons or nodes. The multi-node model and its error back propagation (Back Propagation, BP) algorithm is one of the mature and widely used artificial neural networks and algorithms (Qiu & Li, 2017; Wang, Dong, & Wang, 2017). The BP neural network model consists of three layers, namely, the input layer, the hidden layer, and the output layer (see Figure 1). Among them, the hidden layer can consist of one or more layers and the specific number of layers shall be determined according to data debugging. The model can summarize the law from mass data and use input data to establish the relationship with the output data to solve the nonlinear optimization problem. The advantage is that it can establish the corresponding relationship with the use of the attribute or connotation of the provided data variables themselves to reach the expected value of the data. Meanwhile, this method uses the information highly to prevent the system data identification method from error accumulation in the cumulative and regressive process of the sequence and has good effect on the residual correction of the grey Verhulst model. As a result, the fitting accuracy and prediction accuracy can be obviously improved.