Reference Manual Of Python For Artificial Intelligence On Agriculture

Abstract

The Artificial Intelligence is a very old field of study and has a rich history. Modern AI was formalized by John McCarthy, considered as father of AI. It is a branch of computer science, founded around early 1950’s. Primarily, the term Artificial Intelligence (or AI) refers to a group of technique that enables a computer or a machine to mimic the behaviour of humans in problem solving tasks. Formally, AI is described as “the study of how to make the computers do things at which, at the moment, people are better” (Rich and Knight, 1991; Rich et al., 2009).”The main aim of AI is to program the computer for performing certain tasks in humanly manner such as knowledgebase, reasoning, learning, planning, problem solving etc. The Machine Learning (ML) techniques are the subset of AI which makes the computers/machines/programs the capable of learning and performing tasks without being explicitly programmed. The ML techniques are not just the way of mimicking human behaviour but the way of mimicking how humans learn things. The main characteristics of machine learning is ‘learning from experience’ for solving any kind of problem. The methods of learning can be categorized into three types: (a) supervised learning algorithm is given with labelled data and the desired output whereas (b) unsupervised learning algorithm is given with unlabelled data and identifies the patterns from the input data and (c) reinforcement learning algorithm allows the ML techniques to capture the learnable things on the basis of rewards or reinforcement. Now, the Deep Learning (DL) technique are the advanced version of machine learning algorithms gained much popularity in the area of image recognistion and computer vision. The artificial neural networks (ANNs) clubbed with representation learning are the backbone of the deep learning concepts. These techniques allows a machine to learn patterns in the dataset with multiple levels of abstractions. The DL models are composed of a series of non-linear layers where each of the layer has the capability of transforming the low-level representations into higher-level representations i.e. into a more abstract representations (LeCun et al., 2015). There are several DL algorithms available now-a-days such as Deep Convolutional Neural Networks, Deep Recurrent Neural networks, Long Short-term Memory (LSTM)”networks that are being applied to different areas of engineering, bioinformatics, agriculture, medical science and many more (Fusco et al., 2021).