This (where the outputs are real-valued).In unsupervised learning,

This chapter gives an overview of the di erent learning types and the basic mechanismof machine learning techniques and algorithms.The objective is to understandwhat is machine learning?,Di erent types of machine learning approachesand machine learning algorithms like Support Vector Machine(SVM),Decision Treeclassi er(DT),K-Nearest Neighbor classi er(K-NN),Random Forest classi er(RF),NaveBayes classi er(NB),Ada BoostM2 classi er,Bagg classi er.1.1 IntroductionMachines are the creations of humans.Humans have the capability to understandsituations and behave according to their consciousness. On the contrary machinesdo not have the capability.Machine learning is the eld of computer science thatgives computers the ability to learn without being explicitly programed.Task likeSpam ltering,face recognition,machine translation,speech recognition,data mining,robot motion etc.. are extremely dicult to programing hand.A complex mappingbetween inputs and outputs is usually observed in learning problems.Mapping thelearning methods depends on the kind of data that we have at our disposal.TheFigure.1.1 illustrated the three kinds of learning methods namely,supervised learning,unsupervisedlearning and reinforcement learning.In supervised learning,prede ned class labels are available for all the training examples.This labeled training data is used to build a model, which is used to predict theFigure 1.1: Types of Machine Learningclass of new input data .the training data in supervise learning is encoded as pairsand the output is often manually annotated.the representation of a learning modeldepends on the types of features used to represent the input and the varying degreesof complexity.A good learning model is able to abstract over its experience to detectunderline patterns.the crucial part of a machine learning is the design and test ofthe models.e.g.,spam or ham. The two most common types of supervised learningare classi cation (where the outputs are discrete labels, as in spam ltering) andregression (where the outputs are real-valued).In unsupervised learning, there is no access to any output values,but the collectionof inputs are there .In this learning method try to analyze the underline patternof the there any correlations between features can we cluster our data set ina few groups which behave similarly, and detect outliers.The two most importantexamples are dimension reduction and clustering.In the case of reinforcement learningdirect access to the correct output is not possible.But a measure of how good orbad an output can be assessed(e.g., a robot or controller) .Methods based on machine learning algorithms have been used extensively forvarious applications in the eld of biology 12. These methods have been utilized indiverse domains like genomics, proteomics and systems biology 7. Speci cally, supervisedmachine learning approaches have found immense importance in numerousbioinformatics prediction methods. A brief review of methodologies for prediction ofprotein function with special emphasis on machine learning methods is available16. The aim of the present article is to provide an overview of the machine learningalgorithms as well as application methods based on these algorithms. The basicFigure 1.2: The Basics of Machine Learning Technologymechanisms given in Figure.??machine learning methods are broken into two phases:Training: A model is learned from a collection of training data. Application: Themodel is used to make decisions about some new test data.