Intelligent Data Analysis Techniques
Arvind Kumar Bansal, Javed Iqbal Khan, S. Kaisar Alam in Introduction to Computational Health Informatics, 2019
Artificial neural network is a multilayered connection of perceptrons to solve a complex problem. A perceptron is a signal-processing center that accepts multiple weighted inputs from other perceptrons, and fires “1” to the outgoing connections when the sum of the weighted inputs is greater than the threshold value. A negative bias is also added to a perceptron to cancel any change in the output value due to random noise. The perceptrons in a layer are connected to the perceptrons in the next consecutive layer using weighted edges, and whole ANN is modeled as a directed acyclic graph. ANNs are used to predict the outcome for a set of inputs. In the first phase, a set of samples with known outcomes are selected, and the weights are adjusted continuously using a back-propagation of error between the actual value and the generated output value. The process is repeated until the error is minimal below a threshold.
Outcome Prediction of Oesophago-Gastric Cancer Using Neural Analysis of Pre- and Postoperative Parameters
Raouf N.G. Naguib, Gajanan V. Sherbet in Artificial Neural Networks in Cancer Diagnosis, Prognosis, and Patient Management, 2001
ANNs are parallel information-processing structures that attempt to emulate certain performance characteristics of the biological neural system [21]. ANNs have found an application in several areas of medicine and cancer research in particular [22,23]. Recent work has looked at the role of ANNs in prediction of outcome for patients with colorectal [24], urological [25], and breast cancer [26]. An artificial neural network consists of many processing elements (neurons) which are organised into groups called layers. A typical network consists of a sequence of layers successively connected by full or random connections. There are typically two layers with connection to the outside world: an input layer where data is presented to the network and an output layer which holds the response of the network to a given input. The mathematical model of an artificial neuron and the structure of a generic feed-forward fully interconnected artificial neural network have been described in Chapter 1.
Reduction and Fixation of Sacroiliac joint Dislocation by the Combined Use of S1 Pedicle Screws and an Iliac Rod
Kai-Uwe Lewandrowski, Donald L. Wise, Debra J. Trantolo, Michael J. Yaszemski, Augustus A. White in Advances in Spinal Fusion, 2003
We set out to determine if patient-specific (as opposed to population-specific) outcomes of lumbar fusion could be predicted in advance of treatment [7,8]. We were most interested in patients with chronic low back pain of at least one year duration, who had failed all manner of conservative care. These patients are generally believed to have pain of discogenic origin, be it from annular tears or some other degenerative process. That this diagnosis is difficult to make is understood and that the treatment is controversial is given, but, alas, this is outside the scope of this discussion. Nevertheless, billions of dollars are spent every year in the United States treating this problem, with questionable results. We wanted to find a way to improve outcomes and help patients and surgeons avoid poor outcomes from what is essentially an unpredictable operative procedure. Artificial neural networks were chosen because it was felt they could integrate all the potential variables and do so in a nonlinear manner, free of all operator bias beyond determining the influencing variables.
Learning-based classification of valence emotion from electroencephalography
Published in International Journal of Neuroscience, 2019
Munaza Ramzan, Suma Dawn
Artificial neural network works as a biological neuron. The three important parameters of the biological neuron are synapses, cell body and axon. The neurons are connected with synapses, constantly receiving the signals to reach to the cell body and if the resulting sum of the signals surpasses a certain threshold, a response is sent through the axon. Similarly, a perceptron takes the weighted sum of inputs and based on some activation functions such as sigmoid Gaussian, etc., it transforms the input into output. The weights (Wi) and learning rate (m) are randomly initialized between 0 and 1. For each training instance the activation function is calculated and then the learning rule is applied to find the error between the actual and the predicted output. The weights are updated if the output of the network is not correct otherwise no change in the weight and bias (B). The two-layer feed-forward network, with the sigmoid activation function, was used in this study. It is formulated as:
Effective heart disease prediction with Grey-wolf with Firefly algorithm-differential evolution (GF-DE) for feature selection and weighted ANN classification
Published in Computer Methods in Biomechanics and Biomedical Engineering, 2022
D. Deepika, N. Balaji
HNN (heuristic optimisation-weight of artificial neural network) is a Meta heuristic model designed to increase the performance of ANN (artificial neural network). Artificial neural network technically induce the brain behaviour to improve a system which can manage difficult prediction problems. It is trained by BP (back propagation) algorithm method. The artificial neural network contains several neurons and utilized synaptic weights that connect neurons. Back propagation updates artificial neuron network weights based on calculated errors that found from the previous data based on ANN principles. Because of complexity and application of various steps that makes delay in learning rate with artificial neural network. Back propagation trains artificial neural networks by tuning changing of weights on each trained cycle, consequently the learning range increased slowly. Obviously, the remarkable time needs to train artificial neural network by back propagation algorithm and the importance of developing prediction accuracy of artificial neural networks that motivate to design new model for optimisation of performance in terms of time and accuracy.
Could advances in representation learning in Artificial Intelligence provide the new paradigm for data integration in drug discovery?
Published in Expert Opinion on Drug Discovery, 2019
Vinaya Vijayan, Andrew D. Rouillard, Deepak K. Rajpal, Pankaj Agarwal
Artificial Intelligence (AI) is a branch of computer science that aims to equip machines with human-like intelligence and the ability to learn from ever-changing environments to successfully achieve their goals. For instance, artificial neural networks (ANN) are mathematical models for data processing, born out of AI, designed to mimic the large array of neurons in the brain that enable humans to parse sensory inputs, learn, and make decisions. A basic ANN consists of an input layer, the hidden layer that is a transformed representation of the input data, and the output layer. Recent advances in computing power and algorithmic innovations have led to the development of deep neural networks from ANN’s, which use multiple hidden layers to learn a hierarchy of increasingly abstract but hopefully more meaningful representations of the data [13]. Deep Learning with these deep neural networks has delivered impressive achievements in the last decade with self-driving cars and language translation [13]. A key concept that has become popular in Deep Learning is that of an Autoencoder.
Related Knowledge Centers
- Neural Network
- Neuroplasticity
- Synapse
- Brain
- Connectionism
- Neural Circuit
- Neuron
- Hebbian Theory
- Deep Learning
- Knowledge Representation & Reasoning