Название: Learning Algorithms for Internet of Things: Applying Python Tools to Improve Data Collection Use for System PerformanceАвтор: G.R. Kanagachidambaresan, N. Bharathi
Издательство: Apress
Год: 2024
Страниц: 304
Язык: английский
Формат: pdf
Размер: 10.1 MB
The advent of Internet of Things (IoT) has paved the way for sensing the environment and smartly responding. This can be further improved by enabling intelligence to the system with the support of machine learning and deep learning techniques. This book describes learning algorithms that can be applied to IoT-based, real-time applications and improve the utilization of data collected and the overall performance of the system.
Many societal challenges and problems can be resolved using a better amalgamation of IoT and learning algorithms. "Smartness" is the buzzword that is realized only with the help of learning algorithms. In addition, it supports researchers with code snippets that focus on the implementation and performance of learning algorithms on IoT based applications such as healthcare, agriculture, transportation, etc. These snippets include Python packages such as Scipy, Scikit-learn, Theano, TensorFlow, Keras, PyTorch, and more. Learning Algorithms for Internet of Things provides you with an easier way to understand the purpose and application of learning algorithms on IoT.
Generally, the learning algorithms can be Machine Learning algorithms, Deep Learning algorithms, genetic algorithms, and supporting optimizers. The commonality behind all the learning algorithms is that they extract information from the input training data and apply the gained knowledge to make predictions and identify new input data. The Machine Learning algorithms enable the computers to gain knowledge from the input data automatically. The past data fed as input is used to train the mathematical models in order to predict the future data. The building blocks of Deep Learning algorithms are artificial neural networks, which form the basis for computation and learn the features of the data. As the number of layers increases appropriately in an artificial neural network, the accuracy will increase, and the algorithm will learn the features with fewer resources.
A genetic algorithm (GA) is based on biological evolution and natural selection. These are heuristic search-based algorithms that can be used in Machine Learning, Artificial Intelligence, and optimization techniques. Generally, these algorithms are useful for real-time applications that are complex in nature and require longer time to resolve such as image processing, circuit design in electronics, etc.
Learning algorithms are widely used in almost all domains. The overall procedure to generate a learning algorithm model is to preprocess the dataset, train the model based on the nature of the data with supervised or unsupervised or deep learning algorithms, and then verify the model by testing. These procedure steps are provided in many Python packages. Keras, TensorFlow, SciPy, PyTorch, Theano, Pandas, Matplotlib, Scikit-learn, Seaborn, and OpenCV are the 10 most important Python packages that support learning algorithms, their preprocessing and output prediction, the visualization of results, etc. The Chapter 2 describes these 10 packages.
What you'll Learn:
Supervised algorithms such as Regression and Classification.
Unsupervised algorithms, like K-means clustering, KNN, hierarchical clustering, principal component analysis, and more.
Artificial neural networks for IoT (architecture, feedback, feed-forward, unsupervised).
Convolutional neural networks for IoT (general, LeNet, AlexNet, VGGNet, GoogLeNet, etc.).
Optimization methods, such as gradient descent, stochastic gradient descent, Adagrad, AdaDelta, and IoT optimization.
Who This Book Is For:
Students interested in learning algorithms and their implementations, as well as researchers in IoT looking to extend their work with learning algorithms.
Скачать Learning Algorithms for Internet of Things: Applying Python Tools to Improve Data Collection Use for System Performance