Introduction

Learning is the acquisition and mastery of data over a site through experience. It shouldn’t be only a human thing but appertains to machines too. The world of computing has transformed drastically from an ineffectual mechanical system right into a Herculean automated technique with the appearance of Artificial Intelligence. Data is the fuel that drives this technology; the recent availability of enormous amounts of information has made it the buzzword in technology. Artificial Intelligence, in its simplest form, is to simulate human intelligence into machines for higher decision-making. 

Artificial intelligence (AI) is a branch of computer science that deals with the simulation of human intelligence processes by machines. The term cognitive computing can be used to confer with AI as computer models are deployed to simulate the human pondering process. Any device which recognizes its current environment and optimizes its goal is claimed to be AI enabled. AI might be broadly categorized as weak or strong. The systems which are designed and trained to perform a specific task are often called weak AI, just like the voice activated systems. They will answer a matter or obey a program command, but cannot work without human intervention. Strong AI is a generalized human cognitive ability. It could solve tasks and find solutions without human intervention. Self driving cars are an example of strong AI which uses Computer Vision, Image Recognition and Deep Learning to pilot a vehicle. AI has made its entry into a wide range of industries that profit each businesses and consumers. Healthcare, education, finance, law and manufacturing are just a few of them. Many technologies like Automation, Machine learning, Machine Vision, Natural Language Processing and Robotics incorporate AI.

The drastic increase within the routine work carried out by humans’ calls for the necessity to automation. Precision and accuracy are the following driving terms that demand the invention of intelligent system in contrasted to the manual systems. Decision making and pattern recognition are the compelling tasks that insist on automation as they require unbiased decisive results which might be acquired through intense learning on the historic data of the concerned domain. This might be achieved through Machine Learning, where it’s required of the system that makes predictions to undergo massive training on the past data to make accurate predictions in the longer term.  A few of the popular applications of ML in day by day life include commute time estimations by providing faster routes, estimating the optimal routes and the worth per trip. Its application may be seen in email intelligence performing spam filters, email classifications and making smart replies. In the realm of banking and private finance it’s used to make credit decisions, prevention of fraudulent transactions. It plays a serious role in healthcare and diagnosis, social networking and private assistants like Siri and Cortana. The list is nearly limitless and keeps growing on a regular basis as increasingly fields are employing AI and ML for his or her day by day activities.

True artificial intelligence is many years away, but we’ve a style of AI called Machine Learning today. AI also often called cognitive computing is forked into two cognate techniques, the Machine Learning and the Deep Learning. Machine learning has occupied a substantial space within the research of constructing good and automatic machines. They will recognize patterns in data without being programmed explicitly. Machine learning provides the tools and technologies to learn from the info and more importantly from the changes in the info. Machine learning algorithms have found its place in lots of applications; from the apps that determine the food you select to those that decides in your next movie to observe including the chat bots that book your saloon appointments are just a few of those stunning Machine Learning applications that rock the knowledge technology industry.  Its counterpart the Deep Learning technique has its functionality inspired from the human brain cells and is gaining more popularity. Deep learning is a subset of machine learning which learns in an incremental fashion moving from the low level categories to the high level categories. Deep Learning algorithms provide more accurate results after they are trained with very large amounts of information. Problems are solved using an end to finish fashion which provides them the name as magic box / black box.. Their performances are optimized with using higher end machines. Deep Learning has its functionality inspired from the human brain cells and is gaining more popularity. Deep learning is definitely a subset of machine learning which learns in an incremental fashion moving from the low level categories to the high level categories. Deep Learning is preferred in applications resembling self driving cars, pixel restorations and natural language processing. These applications simply blow our minds but the truth is that absolutely the powers of those technologies are yet to be divulged. This text provides an outline of those technologies encapsulating the speculation behind them together with their applications.

What’s Machine Learning? 

Computers can do only what they’re programmed to do.  This was the story of the past until computers can perform operations and make decisions like human beings. Machine Learning, which is a subset of AI is the technique that permits computers to mimic human beings. The term Machine Learning was invented by Arthur Samuel within the 12 months 1952, when he designed the primary computer program that might learn because it executed. Arthur Samuel was a pioneer of in two most wanted fields, artificial intelligence and computer gaming. In line with him Machine Learning is the “Field of study that offers computers the potential to learn without being explicitly programmed”.

In atypical terms, Machine Learning is a subset of Artificial Intelligence that permits a software to learn by itself from the past experience and use that knowledge to enhance their performance in the longer term works without being programmed explicitly. Consider an example to discover the several flowers based on different attributes like color, shape, smell, petal size etc., In traditional programming all of the tasks are hardcoded with some rules to be followed within the identification process. In machine learning this task might be completed easily by making the machine learn without being programmed. Machines learn from the info provided to them. Data is the fuel which drives the training process. Though the term Machine learning was introduced way back in 1959, the fuel that drives this technology is offered only now. Machine learning requires huge data and computational power which was once a dream is now at our disposal.

Traditional programming Vs Machine Learning:

When computers are employed to perform some tasks as a substitute of human beings, they require to be supplied with some instructions called a pc program. Traditional programming has been in practice for greater than a century. They began within the mid 1800s where a pc program uses the info and runs on a pc system to generate the output. For instance, a historically programmed business evaluation will take the business data and the principles (computer program) as input and can output the business insights by applying the principles to the info. 

Quite the opposite, in Machine learning the info and the outputs also called labels are provided because the input to an algorithm which comes up with a model, as an output.   

For instance, if the shopper demographics and transactions  are fed as input data and use the past customer churn rates because the output data (labels), an algorithm will have the option to construct a model that may predict whether a customer will churn or not. That model known as as a predictive model. Such machine learning models might be used to predict any situation being supplied with the needed historic data. Machine learning techniques are very invaluable ones because they permit the computers to learn recent rules in a high dimensional complex space, that are harder to grasp by the humans.

 Need for Machine Learning:

Machine learning has been around for some time now, but the power to use mathematical calculations mechanically and quickly to large data is now gaining momentum. Machine Learning may be used to automate many tasks, specifically those that may be performed only by humans with their inbred intelligence. This intelligence may be replicated to machines through machine learning. 

Machine learning has found its place in applications just like the self-driving cars, online advice engines like friend recommendations on Facebook and offer suggestions from Amazon, and in detecting cyber frauds. Machine learning is required for problem like image and speech recognition, language translation and sales forecasting, where we cannot write down the fixed rules to be followed for the issue. 

Operations resembling decision making, forecasting, making prediction, providing alerts on deviations, uncovering hidden trends or relationships require diverse, plenty of unstructured and real time data from various artifacts that might be best handled only by machine learning paradigm.

History of Machine Learning

This section discusses concerning the development of machine learning over time. Today we’re witnessing some astounding applications like self driving cars, natural language processing and facial recognition systems making use of ML techniques for his or her processing. All this began within the 12 months 1943, when Warren McCulloch a neurophysiologist together with a mathematician named Walter Pitts authored a paper which threw a lightweight on neurons and its working. They created a model with electrical circuits and thus neural network was born. 

The famous “Turing Test” was created in 1950 by Alan Turing which might ascertain whether the computers had real intelligence. It has to make a human consider that it shouldn’t be a pc but a human as a substitute, to get through the test. Arthur Samuel developed the primary computer program that might learn because it played the sport of checkers within the 12 months 1952. The primary neural network called the perceptron was designed by Frank Rosenblatt within the 12 months 1957. 

The massive shift happened within the Nineties where machine learning moved from being knowledge driven to a knowledge driven technique resulting from the provision of the massive volumes of information. IBM’s Deep Blue, developed in 1997 was the primary machine to defeat the world champion in the sport of chess. Businesses have recognized that the potential for complex calculations might be increased through machine learning.  A few of the latest projects include: Google Brain that was developed in 2012, was a deep neural network that focused on pattern recognition in images and videos. It was later employed to detect objects in You Tube videos. In 2014, Face book created Deep Face which might recognize people similar to how humans do. In 2014, Deep Mind, created a pc program called Alpha Go a board game that defeated knowledgeable Go player. As a consequence of its complexity the sport is claimed to be a really difficult, yet a classical game for artificial intelligence. Scientists Stephen Hawking and Stuart Russel have felt that if AI gains the ability to revamp itself with an intensifying rate, then an unbeatable “intelligence explosion” may result in human extinction. Musk characterizes AI as humanity’s “biggest existential threat.” Open AI is a company created by Elon Musk in 2015 to develop protected and friendly AI that may benefit humanity. Recently, a number of the breakthrough areas in AI are Computer Vision, Natural Language Processing and Reinforcement Learning.

Features of Machine Learning

In recent times technology domain has witnessed an immensely popular topic called Machine Learning. Almost every business is attempting to embrace this technology. Corporations have transformed the way in which by which they carryout business and the longer term seems brighter and promising resulting from the impact of machine learning. A few of the key features of machine learning may include: 

Automation: The capability to automate repetitive tasks and hence increase the business productivity is the largest key factor of machine learning. ML powered paperwork and email automation are getting used by many organizations. Within the financial sector ML makes the accounting work faster, accurate and draws useful insights quickly and simply. Email classification is a classic example of automation, where spam emails are mechanically classified by Gmail into the spam folder. 

Improved customer engagement: Providing a customized experience for purchasers and providing excellent service are very vital for any business to advertise their brand loyalty and to retain long – standing customer relationships. These might be achieved through ML. Creating advice engines which are tailored perfectly to the shopper’s needs and creating chat bots which could simulate human conversations easily by understanding the nuances of conversations and answer questions appropriately. An AVA of Air Asia airline is an example of 1 such chat bots. It’s a virtual assistant that’s powered by AI and responds to customer queries immediately. It could mimic 11 human languages and makes use of natural language understanding technique.

Automated data visualization:  We’re aware that vast data is being generated by businesses, machines and individuals. Businesses generate data from transactions, e-commerce, medical records, financial systems etc. Machines also generate huge amounts of information from satellites, sensors, cameras, computer log files, IoT systems, cameras etc. Individuals generate huge data from social networks, emails, blogs, Web etc. The relationships between the info might be identified easily through visualizations. Identifying patterns and trends in data might be easily done easily through a visible summary of knowledge slightly than going through 1000’s of rows on a spreadsheet. Businesses can acquire invaluable recent insights through data visualizations in-order to extend productivity of their domain through user-friendly automated data visualization platforms provided by machine learning applications. Auto Viz is one such platform that gives automated data visualization tolls to reinforce productivity in businesses.

Accurate data evaluation: The aim of information evaluation is to seek out answers to specific questions that attempt to discover business analytics and business intelligence. Traditional data evaluation involves plenty of trial and error methods, which change into absolutely not possible when working with large amounts of each structured and unstructured data. Data evaluation is an important task which requires huge amounts of time. Machine learning turns out to be useful by offering many algorithms and data driven models that may perfectly handle real time data. 

Business intelligence: Business intelligence refers to streamlined operations of collecting; processing and analyzing of information in a company .Business intelligence applications when powered by AI can scrutinize recent data and recognize the patterns and trends which are relevant to the organization. When machine learning features are combined with big data analytics it could help businesses to seek out solutions to the issues that may help the companies to grow and make more profit. ML has change into probably the most powerful technologies to extend business operations from e-commerce to financial sector to healthcare.  

Languages for Machine Learning

There are various programming languages on the market for machine learning. The selection of the language and the extent of programming desired rely on how machine learning is utilized in an application. The basics of programming, logic, data structures, algorithms and memory management are needed to implement machine learning techniques for any business applications. With this data one can immediately implement machine learning models with the assistance of the varied built-in libraries offered by many programming languages.  There are also many graphical and scripting languages like Orange, Big ML, Weka and others allows to implement ML algorithms without being hardcoded;  all that you just require is only a fundamental knowledge about programming.

There isn’t a single programming language that might be called because the ‘best’ for machine learning. Each of them is nice where they’re applied. Some may prefer to make use of Python for NLP applications, while others may prefer R or Python for sentiment evaluation application and a few use Java for ML applications referring to security and threat detection.  Five different languages which are best fitted to ML programming is listed below.

Python:

 Nearly 8. 2 million developers are using Python for coding around the globe. The annual rating by the IEEE Spectrum, Python was chosen as the most well-liked programming language. It also seen that the Stack overflow trends in programming languages show that Python is rising for the past five years. It has an in depth collection of packages and libraries for Machine Learning. Any user with the fundamental knowledge of Python programming can use these libraries instantly without much difficulty.

To work with text data, packages like NLTK, SciKit and Numpy comes handy.  OpenCV and Sci-Kit image may be used to process images. One can use Librosa while working with audio data. In implementing deep learning applications, TensorFlow, Keras and PyTorch are available in as a life saver. Sci-Kit-learn may be used for implementing primitive machine learning algorithms and Sci-Py for performing scientific calculations. Packages like Matplotlib, Sci-Kit and Seaborn are best fitted to best data visualizations. 

R:

R is a wonderful programming language for machine learning applications using statistical data. R is filled with a wide range of tools to coach and evaluate machine learning models to make accurate future predictions. R is an open source programming language and really cost effective. It is extremely flexible and cross-platform compatible. It has a broad spectrum of techniques for data sampling, data evaluation, model evaluation and data visualization operations. The excellent list of packages include MICE which is used for handling missing values, CARET to perform classification an regression problems, PARTY and rpart to create partitions in data, random FOREST for crating decision trees, tidyr and dplyr are used for data manipulation, ggplot for creating data visualizations, Rmarkdown and Shiny to perceive insights through the creation of reports. 

Java and JavaScript:

Java is picking up more attention in machine learning from the engineers who come from java background. Many of the open source tools like Hadoop and Spark which are used for large data processing are written in Java. It has a wide range of third party libraries like JavaML to implement machine learning algorithms. Arbiter Java is used for hyper parameter tuning in ML. The others are Deeplearning4J and Neuroph that are utilized in deep learning applications. Scalability of Java is an awesome lift to ML algorithms which enables the creation of complex and big applications. Java virtual machines are an added advantage to create code on multiple platforms.

Julia:

Julia is a general purpose programming language that’s able to performing complex numerical evaluation and computational science. It’s specifically designed to perform mathematical and scientific operations in machine learning algorithms. Julia code is executed at high speed and doesn’t require any optimization techniques to handle problems referring to performance. Has a wide range of tools like TensorFlow, MLBase.jl, Flux.jl, SciKitlearn.jl. It supports all kinds of hardware including TPU’s and GPU’s. Tech giants like Apple and Oracle are emplying Julia for his or her machine learning applications.

Lisp:

LIST (List Processing) is the second oldest programming language which is getting used still. It was developed for AI-centric applications. LISP is utilized in inductive logic programming and machine learning. ELIZA, the primary AI chat bot was developed using LISP. Many machine learning applications like chatbots eCommerce are developed using LISP. It provides quick prototyping capabilities, does automatic garbage collection, offers dynamic object creation and provides lot of flexibility in operations.

Varieties of Machine Learning

At a high-level machine learning is defined because the study of teaching a pc program or an algorithm to mechanically improve on a selected task. From the research point, it might be viewed through the attention of theoretical and mathematical modeling, concerning the working of the whole process. It’s interesting to learn and understand about the different sorts of machine learning in a world that’s drenched in artificial intelligence and machine learning. From the angle of a pc user, this may be seen because the understanding of the varieties of machine learning and the way they might reveal themselves in various applications. And from the practitioner’s perspective it’s needed to know the varieties of machine learning for creating these applications for any given task. 

Supervised Learning:

Supervised learning is the category of problems that uses a model to learn the mapping between the input variables and the goal variable. Applications consisting of the training data describing the varied input variables andthe goal variable are often called supervised learning tasks.

 Let the set of input variable be (x) and the goal variable be (y). A supervised learning algorithm tries to learn a hypothetical function which is a mapping given by the expression y=f(x), which is a function of x. 

The educational process here is monitored or supervised. Since we already know the output the algorithm is corrected every time it makes a prediction, to optimize the outcomes. Models are fit on training data which consists of each the input and the output variable after which it’s used to make predictions on test data. Only the inputs are provided in the course of the test phase and the outputs produced by the model are compared with the kept back goal variables and is used to estimate the performance of the model.

There are mainly two varieties of supervised problems: Classification – which involves prediction of a category label and Regression – that involves the prediction of a numerical value.

The MINST handwritten digits data set may be seen for example of classification task. The inputs are the photographs of handwritten digits, and the output is a category label which identifies the digits within the range 0 to 9 into different classes. 

The Boston house price data set might be seen for example of Regression problem where the inputs are the features of the home, and the output is the worth of a house in dollars, which is a numerical value.  

Unsupervised Learning:

In an unsupervised learning problem the model tries to learn by itself and recognize patterns and extract the relationships amongst the info. As in case of a supervised learning there is no such thing as a supervisor or a teacher to drive the model. Unsupervised learning operates only on the input variables. There are not any goal variables to guide the training process. The goal here is to interpret the underlying patterns in the info as a way to obtain more proficiency over the underlying data. 

There are two most important categories in unsupervised learning; they’re clustering – where the duty is to seek out out the several groups in the info. And the following is Density Estimation – which tries to consolidate the distribution of information.   These operations are performed to know the patterns in the info. Visualization and Projection may additionally be regarded as unsupervised as they struggle to supply more insight into the info. Visualization involves creating plots and graphs on the info and Projection is involved with the dimensionality reduction of the info.

Reinforcement Learning:

Reinforcement learning is type a of problem where there’s an agent and the agent is working in an environment based on the feedback or reward given to the agent by the environment by which it is working. The rewards might be either positive or negative. The agent then proceeds within the environment based on the rewards gained.    

The reinforcement agent determines the steps to perform a specific task. There isn’t a fixed training dataset here and the machine learns by itself. 

Playing a game is a classic example of a reinforcement problem, where the agent’s goal is to accumulate a high rating. It makes the successive moves in the sport based on the feedback given by the environment which could also be when it comes to rewards or a penalization. Reinforcement learning has shown tremendous ends in Google’s AplhaGo of Google which defeated the world’s primary Go player.

Machine Learning Algorithms

There are a number of machine learning algorithms available and it is vitally difficult and time consuming to pick essentially the most appropriate one for the issue at hand. These algorithms may be grouped in to 2 categories. Firstly, they may be grouped based on their learning pattern and secondly by their similarity of their function.

Based on their learning style they may be divided into three types:

1. Supervised Learning Algorithms: The training data is provided together with the label which guides the training process. The model is trained until the specified level of accuracy is attained with the training data. Examples of such problems are classification and regression. Examples of algorithms used include Logistic Regression, Nearest Neighbor, Naive Bayes, Decision Trees, Linear Regression, Support Vector Machines (SVM), Neural Networks.

2. Unsupervised Learning Algorithms: Input data shouldn’t be labeled and doesn’t include a label. The model is ready by identifying the patterns present within the input data. Examples of such problems include clustering, dimensionality reduction and association rule learning. List of algorithms used for these style of problems include Apriori algorithm and K-Means and Association Rules 
3. Semi-Supervised Learning Algorithms: The price to label the info is kind of expensive because it requires the knowledge of expert human experts. The input data is combination of each labeled and unlabelled data. The model makes the predictions by learning the underlying patterns on their very own. It’s a combination of each classification and clustering problems. 

Based on the similarity of function the algorithms may be grouped into the next:

1. Regression Algorithms: Regression is a process that is worried with identifying the connection between the goal output variables and the input features to make predictions concerning the recent data.  Top six Regression algorithms are: Easy Linear Regression, Lasso Regression, Logistic regression, Multivariate Regression algorithm, Multiple Regression Algorithm.

2. Instance based Algorithms: These belong to the family of learning that measures recent instances of the issue with those within the training data to seek out out a best match and makes a prediction accordingly. The highest instance based algorithms are: k-Nearest Neighbor, Learning Vector Quantization, Self-Organizing Map, Locally Weighted Learning, and Support Vector Machines. 
3. Regularization: Regularization refers back to the strategy of regularizing the training process from a specific set of features. It normalizes and moderates. The weights attached to the features are normalized which prevents in certain features dominating the prediction process. This system helps to forestall the issue of overfitting in machine learning. The varied regularization algorithms are Ridge Regression, Least Absolute Shrinkage and Selection Operator (LASSO) and Least-Angle Regression (LARS).

4. Decision Tree Algorithms: These methods construct tree based model constructed on the choices made by examining the values of the attributes. Decision trees are used for each classification and regression problems. A few of the well-known decision tree algorithms are: Classification and Regression Tree, C4.5 and C5.0, Conditional Decision Trees, Chi-squared Automatic Interaction Detection and Decision Stump.

5. Bayesian Algorithms: These algorithms apply the Bayes theorem for the classification and regression problems. They include Naive Bayes, Gaussian Naive Bayes, Multinomial Naive Bayes, Bayesian Belief Network, Bayesian Network and Averaged One-Dependence Estimators.

6. Clustering Algorithms: Clustering algorithms involves the grouping of information points into clusters. All the info points which are in the identical group share similar properties and, data points in numerous groups have highly dissimilar properties. Clustering is an unsupervised learning approach and is generally used for statistical data evaluation in lots of fields. Algorithms like k-Means, k-Medians, Expectation Maximisation, Hierarchical Clustering, Density-Based Spatial Clustering of Applications with Noise fall under this category.

7. Association Rule Learning Algorithms: Association rule learning is a rule-based learning method for identifying the relationships between variables in a really large dataset. Association Rule learning is employed predominantly in market basket evaluation. The preferred algorithms are: the Apriori algorithm and the Eclat algorithm.

8. Artificial Neural Network Algorithms: Artificial neural network algorithms relies find its base from the biological neurons within the human brain. They belong to the category of complex pattern matching and prediction process in classification and regression problems. A few of the popular artificial neural network algorithms are: Perceptron, Multilayer Perceptrons, Stochastic Gradient Descent, Back-Propagation, , Hopfield Network, and Radial Basis Function Network.  

9. Deep Learning Algorithms: These are modernized versions of artificial neural network, that may handle very large and complicated databases of labeled data. Deep learning algorithms are tailored to handle text, image, audio and video data. Deep learning uses self-taught learning constructs with many hidden layers, to handle big data and provides more powerful computational resources. The preferred deep learning algorithms are: A few of the popular deep learning ms include Convolutional Neural Network, Recurrent Neural Networks, Deep Boltzmann Machine, Auto-Encoders Deep Belief Networks and Long Short-Term Memory Networks. 

10. Dimensionality Reduction Algorithms: Dimensionality Reduction algorithms exploit the intrinsic structure of information in an unsupervised manner to precise data using reduced information set. They convert a high dimensional data right into a lower dimension which might be utilized in supervised learning methods like classification and regression. A few of the well-known dimensionality reduction algorithms include Principal Component Evaluation, Principal Component Regressio, Linear Discriminant Evaluation, Quadratic Discriminant Evaluation, Mixture Discriminant Evaluation, Flexible Discriminant Evaluation and Sammon Mapping.

11. Ensemble Algorithms: Ensemble methods are models made up of varied weaker models which are trained individually and the person predictions of the models are combined using some method to get the ultimate overall prediction. The standard of the output is dependent upon the strategy chosen to mix the person results. A few of the popular methods are: Random Forest, Boosting, Bootstrapped Aggregation, AdaBoost, Stacked Generalization, Gradient Boosting Machines, Gradient Boosted Regression Trees and Weighted Average.

Machine Learning Life Cycle

Machine learning gives the power to computers to learn mechanically without having the necessity to program them explicitly. The machine learning process comprises of several stages to design, develop and deploy prime quality models. Machine Learning Life Cycle comprises of the next steps

1. Data collection
2. Data Preparation 
3. Data Wrangling
4. Data Evaluation
5. Model Training
6. Model Testing
7. Deployment of the Model

1. Data Collection: That is the very first step in making a machine learning model. The most important purpose of this step is to discover and gather all the info which are relevant to the issue. Data might be collected from various sources like files, database, web, IoT devices, and the list is ever growing. The efficiency of the output will depend directly on the standard of information gathered. So utmost care ought to be taken in gathering large volume of quality data. 
2. Data Preparation: The collected data are organized and put in a single place or further processing. Data exploration is part of this step, where the characteristics, nature, format and the standard of the info are being accessed. This includes creating pie charts, bar charts, histogram, skewness etc. data exploration provides useful insight on the info and is useful in solving of 75% of the issue.

3. Data Wrangling: In Data Wrangling the raw data is cleaned and converted right into a useful format. The common technique applied to make essentially the most out of the collected data are:

1. Missing value check and missing value imputation
2. Removing unwanted data and Null values
3. Optimizing the info based on the domain of interest
4. Detecting and removing outliers
5. Reducing the dimension of the info
6. Balancing the info, Under-Sampling and Over-Sampling.
7. Removal of duplicate records

4. Data Evaluation: This step is worried with the feature selection and model selection process. The predictive power of the independent variables in relation to the dependent variable is estimated. Only those variables which are useful to the model is chosen. Next the suitable machine learning technique like classification, regression, clustering, association, etc is chosen and the model is built using the info. 

5. Model Training: Training is an important step in machine learning, because the model tries to know the varied patterns, features and the principles from the underlying data. Data is split into training data and testing data. The model is trained on the training data until its performance reaches a suitable level.

6. Model Testing: After training the model it’s put under testing to guage its performance on the unseen test data. The accuracy of prediction and the performance of the model may be measured using various measures like confusion matrix, precision and recall, Sensitivity and specificity, Area under the curve, F1 rating, R square, gini values etc.

7. Deployment: That is the ultimate step within the machine learning life cycle, and we deploy the model constructed in the true world system. Before deployment the model is pickled that’s it needs to be converted right into a platform independent executable form. The pickled model may be deployed using Rest API or Micro-Services. 

Deep Learning

Deep learning is a subset of machine learning that follows the functionality of the neurons within the human brain. The deep learning network is made up of multiple neurons interconnected with one another in layers. The neural network has many deep layers that enable the training process. The deep learning neural network is made up of an input layer, an output layer and multiple hidden layers that make up the entire network. The processing happens through the connections that contain the input data, the pre-assigned weights and the activation function which decides the trail for the flow of control through the network. The network operates on huge volume of information and propagates them thorough each layer by learning complex features at each level. If the consequence of the model shouldn’t be as expected then the weights are adjusted and the method repeats again until the will consequence is achieved.

Deep neural network can learn the features mechanically without being programmed explicitly. Each layer depicts a deeper level of knowledge. The deep learning model follows a hierarchy of data represented in each of the layers. A neural network with five layers will learn greater than a neural network with three layers. The educational in a neural network occurs in two steps. In step one, a nonlinear transformation is applied to the input and a statistical model is created. Throughout the second step, the created model is improved with the assistance of a mathematical model called as derivative. These two steps are repeated by the neural network 1000’s of times until it reaches the specified level of accuracy. The repetition of those two steps is often called iteration. 

The neural network that has just one hidden layer is often called a shallow network and the neural network that has multiple hidden layers is often called deep neural network.

Varieties of neural networks:

There are several types of neural networks available for several types of processes. Probably the most commonly used types are discussed here.

1. Perceptron: The perceptron is a single-layered neural network that comprises only an input layer and an output layer. There are not any hidden layers. The activation function used here is the sigmoid function.

2. Feed forward:  The feed forward neural network is the only type of neural network where the knowledge flows only in a single direction. There are not any cycles in the trail of the neural network. Every node in a layer is connected to all of the nodes in the following layer. So all of the nodes are fully connected and there are not any back loops.

3. Recurrent Neural Networks: Recurrent Neural Networks saves the output of the network in its memory and feeds it back to the network to assist in the prediction of the output. The network is made up of two different layers. The primary is a feed forward neural network and the second is a recurrent neural network where the previous network values and states are remembered in a memory. If a mistaken prediction is made then the training rate is used to steadily move towards making the proper prediction through back propagation. 

4. Convolutional Neural Network: Convolutional Neural Networks are used where it’s needed to extract useful information from unstructured data. Propagation of signa is uni-directional in a CNN. The primary layer is convolutional layer which is followed by a pooling, followed by multiple convolutional and pooling layers. The output of those layers is fed into a completely connected layer and a softmax that performs the classification process. The neurons in a CNN have learnable weights and biases. Convolution uses the nonlinear RELU activation function. CNNs are utilized in signal and image processing applications. 

5. Reinforcement Learning: In reinforcement learning the agent that operates in a fancy and unsure environment learns by a trial and error method. The agent is rewarded or punished virtually in consequence of its actions, and helps in refining the output produced. The goal is to maximise the whole variety of rewards received by the agent. The model learns by itself to maximise the rewards. Google’s DeepMind and Self drivig cars are examples of applications where reinforcement learning is leveraged. 

Difference Between Machine Learning And Deep Learning

Deep learning is a subset of machine learning. The machine learning models change into higher progressively as they learn their functions with some guidance. If the predictions will not be correct then an authority has to make the adjustments to the model.  In deep learning the model itself is able to identifying whether the predictions are correct or not.

• Functioning: Deep learning takes the info because the input and tries to make intelligent decisions mechanically using the staked layers of artificial neural network. Machine learning takes the input data, parses it and gets trained on the info. It tries to make decisions on the info based on what it has learnt in the course of the training phase.

• Feature extraction: Deep learning extracts the relevant features from the input data. It mechanically extracts the features in a hierarchical manner. The features are learnt in a layer clever manner. It learns the low-level features initially and because it moves down the network it tries to learn the more specific features. Whereas machine learning models requires features which are hand-picked from the dataset. These features are provided because the input to the model to do the prediction.

• Data dependency: Deep learning models require huge volumes of information as they do the feature extraction process on their very own. But a machine learning model works perfectly well with smaller datasets. The depth of the network in a deep learning model increases with the info and hence the complexity of the deep learning model also increases. The next diagram shows that the performance of the deep learning model increases with increased data, however the machine learning models flattens the curve after a certain period.

• Computational Power: Deep learning networks are highly depending on huge data which requires the support of GPUs slightly than the traditional CPUs.  GPUs can maximize the processing of deep learning models as they will process multiple computations at the identical time. The high memory bandwidth in GPUs makes them suitable for deep learning models. Then again machine learning models may be implemented on CPUs. 

• Execution time: Normally deep learning algorithms take an extended time to coach resulting from the big variety of parameters involved. The ResNet architecture which is an example of deep learning algorithm takes almost two weeks to coach from the scratch. But machine learning algorithms takes less time to coach (jiffy to just a few hours). This is totally reversed with respect to the testing time. Deep learning algorithms take lesser time to run. 

• Interpretability: It is less complicated to interpret machine learning algorithms and understand what’s being done at each step and why it’s being done. But deep learning algorithms are often called black boxes as one really doesn’t know what is occurring on the inside the deep learning architecture. Which neurons are activated and the way much they contribute to the output. So interpretation of machine learning models is far easier than the deep learning models.

Applications of Machine Learning

• Traffic Assistants: All of us use traffic assistants after we travel. Google Maps turns out to be useful to present us the routes to our destination and likewise shows us the routes with less traffic. Everyone who uses the maps are providing their location, route taken and their speed of driving to Google maps. These details concerning the traffic are collected by Google Maps and it tries to predict the traffic in your route and tries to regulate your route accordingly.

• Social media: Probably the most common application of machine learning might be seen in automatic friend tagging and friend suggestions. Facebook uses Deep Face to do Image recognition and Face detection in digital images. 

• Product Suggestion: Whenever you flick through Amazon for a specific product but don’t purchase them, then the following day once you open up YouTube or Facebook then you definitely get to see ads referring to it. Your search history is being tracked by Google and it recommends products based in your search history. That is an application of machine learning technique.

• Personal Assistants: Personal assistants help to find useful information. The input to a private assistant might be either through voice or text. There isn’t a one who could say that they don’t learn about Siri and Alexa. Personal assistants may also help in answering phone calls, scheduling meeting, taking notes, sending emails, etc.

• Sentiment Evaluation: It’s an actual time machine learning application that may understand the opinion of individuals. Its application may be viewed in review based web sites and in decision making applications. 

• Language Translation: Translating languages is not any more a difficult task as there’s a hand stuffed with language translators available now. Google’s GNMT is an efficient neural machine translation tool that may access 1000’s of dictionaries and languages to supply an accurate translation of sentences or words using the Natural Language Processing technology.

• Online Fraud Detection: ML algorithms can learn from historical fraud patterns and recognize fraud transaction in the longer term.ML algorithms have proved to be more efficient than humans within the speed of knowledge processing. Fraud detection system powered by ML can find frauds that humans fail to detect.  

• Healthcare services: AI is becoming the longer term of healthcare industry. AI plays a key role in clinical decision making thereby enabling early detection of diseases and to customize treatments for patients. PathAI which uses machine learning is utilized by pathologists to diagnose diseases accurately. Quantitative Insights is AI enabled software that improves the speed and accuracy within the diagnosis of breast cancer.  It provides higher results for patients through improved diagnosis by radiologists.

Applications of Deep Learning

• Self-driving cars: Autonomous driving cars are enabled by deep learning technology. Research can be being done on the Ai Labs to integrate features like food delivery into driverless cars. Data is collected from sensors, cameras and geo mapping helps to create more sophisticated models that may travel seamlessly through traffic. 
• Fraud news detection: Detecting fraud news may be very vital in today’s world. Web has change into the source of all types of reports each real and pretend. Attempting to discover fake news is a really difficult task. With the assistance of deep learning we will detect fake news and take away it from the news feeds. 
• Natural Language Processing: Trying to know the syntaxes, semantics, tones or nuances of a language is a really hard and complicated task for humans. Machines might be trained to discover the nuances of a language and to border responses accordingly with the assistance of Natural Language Processing technique. Deep learning is gaining popularity in applications like classifying text, twitter evaluation, language modeling, sentiment evaluation etc, which employs natural language processing. 
• Virtual Assistants: Virtual assistants are using deep learning techniques to have an in depth knowledge concerning the subjects right from people’s dining out preferences to their favorite songs. Virtual assistants try to know the languages spoken and take a look at to perform the tasks. Google has been working on this technology for a few years called Google duplex which uses natural language understanding, deep learning and text-to–speech to assist people book appointments anywhere in the midst of the week. And once the assistant is completed with the job it gives you a confirmation notification that your appointment has been taken care of. The calls don’t go as expected however the assistant understands the context to nuance and handles the conversation gracefully. 
• Visual Recognition: Going through old photographs might be nostalgic, but trying to find a specific photo could change into a tedious process because it involves sorting, and segregation which is time consuming. Deep learning can now be applied o images to sort them based on locations in the images, combination of peoples, based on some events or dates. Searching the images is not any more a tedious and complicated. Vision AI draws insights from images within the cloud with AutoML Vision or pretrained Vision API models to discover text, understand emotions in images. 
• Coloring of Black and White images:  Coloring a black and white image is like a baby’s play with the assistance of Computer Vision algorithms that use deep learning techniques to bring concerning the life in the images by coloring them with the proper tones of color. The Colourful Image Colorization micro-services is an algorithm using computer vision technique and deep learning algorithms which are trained on the Imagenet database to paint black and white images.
• Adding Sounds to Silent Movies:  AI can now create realistic sound tracks for silent videos. CNNs and recurrent neural networks are employed to perform feature extraction and the prediction process. Research have shown that these algorithms which have learned to predict sound can produce higher sound effects for old movies and help robots understand the objects of their surroundings.
• Image to Language Translation: That is one other interesting application of deep learning. The Google translate app can mechanically translate images into real time language of selection. The deep learning network reads the image and translates the text into the needed language.
• Pixel Restoration: The researchers in Google Brain have trained a Deep Learning network that takes a really low resolution image of  an individual faces and predicts the person’s face through it. This method is often called Pixel Recursive Super Resolution. This method enhances the resolution of photos by identifying the outstanding features that’s simply enough for identifying the personality of the person. 

Conclusion

This chapter has discovered the applications of machine learning and deep learning to present a clearer idea concerning the current and future capabilities of Artificial Intelligence. It’s predicted that many applications of Artificial Intelligence will affect our lives within the near future. Predictive analytics and artificial intelligence are going to play a fundamental role in  the longer term in content creation and likewise within the software  development. Actually, the very fact is that they are already making an impact. Inside the following few years, AI development tools, libraries, and languages will change into the universally accepted standard components of each software development toolkit that you would be able to name. The technology of artificial intelligence will change into the longer term in all of the domains including health, business, environment, public safety and security.

References

[1] Aditya Sharma(2018), “Differences Between Machine Learning & Deep Learning”  

[2] Kislay Keshari(2020), “Top 10 Applications of Machine Learning : Machine Learning Applications in Each day Life” 

[3] Brett Grossfeld(2020), “Deep learning vs machine learning: an easy option to understand the difference”    

[4] By Nikita Duggal(2020), “Real-World Machine Learning Applications That Will Blow Your Mind”    

[5] P. P. Shinde and S. Shah, “A Review of Machine Learning and Deep Learning Applications,” 2018 Fourth International Conference on Computing Communication Control and Automation (ICCUBEA), Pune, India, 2018, pp. 1-6    

[6] https://www.javatpoint.com/machine-learning-life-cycle                                 

[7] https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-112a47a429d0

[8]  Dan Shewan(2019), “10 Corporations Using Machine Learning in Cool Ways”

[9]  Marina Chatterjee(2019), “Top 20 Applications of Deep Learning in 2020 Across Industries

[10] A Tour of Machine Learning Algorithms by Jason Brownlee in Machine Learning Algorithms

[11]  Jaderberg, Max, et al. “Spatial Transformer Networks.” In Advances in neural information processing systems (2015): 2017-2025.

[12] Van Veen, F. & Leijnen, S. (2019). The Neural Network Zoo. Retrieved from https://www.asimovinstitute.org/neural-network-zoo

[13] Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton,  ImageNet Classification with Deep Convolutional Neural Networks, [pdf], 2012

[14] Yadav, Neha, Anupam, Kumar, Manoj, An Introduction to Neural Networks for Differential Equations (ISBN: 978-94-017-9815-0)

[15] Hugo Mayo, Hashan Punchihewa, Julie Emile, Jackson Morrison History of Machine Learning, 2018

[16] Pedro Domingos , 2012, Tapping into the “folk knowledge” needed to advance machine learning applications. by A Few Useful, doi:10.1145/2347736.2347755

[17] Alex Smola and S.V.N. Vishwanathan, Introduction to Machine Learning, Cambridge University Press 2008

[18] Antonio Guili and Sujit Pal, Deep Learning with Keras: Implementing deep learning models and neural networks with the ability of Python, Release 12 months: 2017; Packt Publishing Ltd.

[19] AurÈlien GÈron ,Hands-On Machine Learning with Scikit-Learn and Tensor Flow: Concepts, Tools, and Techniques to Construct Intelligent Systems, Release 12 months: 2017. O’Reilly

[20] Best language for Machine Learning: Which Programming Language to Learn, August 31, 2020, Springboard India.