We use cookies to give you the best experience possible. By continuing we’ll assume you’re on board with our cookie policy

Check Writers' Offers

What's Your Topic?

Hire a Professional Writer Now

The input space is limited by 250 symbols

What's Your Deadline?

Choose 3 Hours or More.
Back
2/4 steps

How Many Pages?

Back
3/4 steps

Sign Up and Get Writers' Offers

"You must agree to out terms of services and privacy policy"
Back
Get Offer

Machine Learning

Paper type: Essay
Pages: 8 (1780 words)
Categories: Learning
Downloads: 35
Views: 6

Abstract

Machine Learning (ML) used in daily life. ML can use data and use it for self-learning. It is widely used in many fields like finance, agriculture, education, and security, etc. This paper discusses the prospective of applying machine learning skills in several segments. ML categorized in mainly four Learning processes. Supervised learning which has labelled data were unsupervised learning includes unlabelled data. Semi-supervised learning is a grouping of supervised and unsupervised learning. The reinforcement learning algorithm is learning by getting an opinion on the outcome of adjusting some factors.

We can study in detail with different techniques and different methods and check which algorithm is more accurate in less runtime. This paper summarizes some applications of machine learning such as prediction, disease detection, fraud detection and more. This paper supports dropping the research gap for ML applications.

Introduction

Machine Learning (ML) can get access to data and use it for self-learning. The main motto stands to allow the processors to learn repeatedly without humanoid intercession or else collaboration and adjust activities according to it.

ML Model consists mainly of four steps: environment, learning, repository and execute [1]. Environment means to acquire outside information. Quality of information affects directly on learning to understand whether easy or clutter. Learning means to get the knowledge from information and that knowledge is adding to the repository which stores many general principles that guide the implementation activities. Thomas H. Davenport transcribes in the Wall Street Journal, “Humans can typically create one or two good models a week (while) machine learning can create thousands of models a week”[2].

Supervised learning

It recognizes a labelled data set and separating in two types: classification and regression [3]. It is widely cast-off in different-different tenders, like object recognition, spam detection and speech recognition [6]. Supervised learning, if applied with proper information, the error rate can compact tens to the lowest fault rate assured [3].  Table 1 describes the tasks performed using supervised learning, the algorithms used for performing the tasks and application domains.

Unsupervised Learning

Unsupervised Learning assets we have a gathering of unlabelled data, we need to analyse and find patterns inside it. Unsupervised learning has two vital illustrations; clustering and dimension reduction [7]. Classic clustering algorithms contain spectrum clustering, hierarchical clustering, k means, Dirichlet process and fuzzy clustering [3, 4]. Table 2 displays algorithms cast off to segment text topics, propose things and find data outliers [5].

Data aggregation

Reinforcement Learning

The algorithm studies by getting a response on the result of transforming some limits, e.g. modulation and power [6]. This machine is training to take exact decisions based on the business need to maximize efficiency. It is using in gaming and navigation and usually, to address applications for instance economics, robotics, inventory management, where the aim is to learn a policy, i.e., planning between states of the environment into movements to accomplished, while openly cooperating with the situation [6, 7]. Table 3 shows the algorithm task and technique used in various applications.

It is a mixture of supervised and unsupervised learning. Self-training exists the oldest form of semi-supervised learning [8]. Example is recognizing a person’s face on a webcam. It consists of methods such as classification, regression, and prediction [5]  The methods can  separated in four classes: i) techniques based on generative models  ii) techniques based on the supposition that the decision boundary should lie in a low-density region iii) graph-based methods  iv) two-step methods The first phase is an unsupervised learning stage to modify the facts representation or build a novel kernel; then a supervised learning stage depend on the novel demonstration or kernel [6].

Application

ML has various applications in diverse fields:

Agriculture

Now Indian agronomy has developed a significant source of financial improvement. The agriculture trades started exploring innovative approaches to grow food assembly because of deviations in climate and growing populace. This makes researchers find new well-organized and valuable skills for great efficiency. Agriculturalists can accumulate records by using precision agriculture. Precision agriculture is a method, which offers advanced skills to increase farm output. By exploiting Precision agriculture it is conceivable to create financial evolution in agronomy and moreover it is cast-off in several applications like weed detection, pest detection in plants, yield growth of plant and crop ailment detection, etc. An agronomist applies pesticides to control pests, avoid sicknesses and to growth crop produce. The ailments in the crop are generating a problem of less production and financial losses to agriculturalists as well as agronomic businesses. So identification of illness and its severity founded to become essential. Table 4 shows different-different techniques are using detect the plant diseases and accuracy of each.

Healthcare

The healthcare arena facing supplementary difficulties and it is becoming more costly. Some ML processes cast-off to fix them, ML methods apply for an estimate the several ailments like thyroid, cardiovascular, heart, diabetic, cancer and hepatitis disease. Since the previous learning, recognized that SVM provides 96.40% accurateness for the breast cancer analysis, CART provides 79% correctness for a discerning of diabetic illness, naive Bayes provides 86% exactness for the diagnosis of heart sickness [4]. Table 5 display the various techniques to detect the particular disease with its accuracy, also mention the dataset collect from which place.

Finance

ML and artificial intelligence in economic business [28] grasps a lot of possibilities, this embrace banks, insurance corporations, credit unions, investment funds, specific fund managers, accountancy establishments, credit card companies, customer finance businesses and stock brokerages [27]. Fig.3 shows ML techniques are using for the finance segments.

Diagnostic Analytics

It is utilizing for knowing the origin cause of a marvel, a scheme success otherwise failure. It comprises all arrangements of economic threat valuation. A real-world execution is an instrument “Halo” via Pricewaterhouse Cooper. Halo workings on accounting papers and books [29]. Its ability to ensign higher-threat transactions. This is not an online actual solution then the exertion is individual to detect and show the financial threats. Accounting is about hoary transaction histories, and that is what a device like a halo attempts to extract data from a user’s systems and define to fetch forth high-threat transactions with their designs and movements. It also provides evidence on development inadequacies in the accounting development, similarly, removing matching accounting journal admissions, etc. This is beneficial for a huge financial association owing to the high-capacity of transactions that occur through a specific time duration in several sections [27].

Predictive Analytics

It is used to show the result of an event in the imminent. For instance, the success otherwise failure of economic business, financial budget overrun or else budget slippage for an institute. For example, a bank or an enterprise. There are different parameters in the ML model code proceeds hooked on account for inquiring before deriving the conclusion. The intention is to offer intelligence to the topmost managing direction that the financial side of the industry. It doesn’t crack the problem however, it simply says or performances a threatening indicator of failure that deviates since the strategy for impending connected to finances. After applying these analytics, we acquire lots of arenas of investment that originate into production, for instance, resolution funding systems, supervisory finances, and rate accountancy. Predictive costing testimony is found on historical cost accountancy documents.it helps enormous economical industries stay on a trajectory with their total budgets and make sure the correct standards are elevated if the real price is predicting to grow the effect of estimation by the ML model. In managing accounting or finances, it recognized in the theoretical groups, the attention on what-if situations [27]. ML models established on historical facts and smearing them is evaluate on the ideologies of managing to account.

Prescriptive Analytics

The greatest value obtains by constructing the prediction model based on predictive analytics and advice for future actions. It uses ML to build the imminent plan built on historical disappointments in fiscal transactions in parallel situations, for example, unsuccessful financial connections, and it provides explicit events that necessary to send in order for the financial operation to achievement. In the financial segment, it is now correctly able to do course changes even already the failures occur. The prescriptive part starts when there is correct prediction completed then the activities that want to take even before this ensues originate later that [27].

Education

ML castoff in education consumes ample emphasis recently. For expecting learner’s presentation, generally used model is Na?ve Bayes, Artificial Neural Network, Support Vector Machine, instance-based learning, clustering, Decision Tree and Classification Tree [30]. Various documents are castoff to analyse the learner’s presentation, for example, demographic aspect and preface of the earlier semester, examination mark, and pre-university. GLM (Generalized Linear Model) is the extended context of linear regression. The awareness of calculating a hyperplane to cut the defective utility [31]. SVM is commonly used for categorization to forecast the connection of information and data mining. It is established on a geometrical explanation. The procedure examines for optimum separating surface, for example, hyperplane [32]. The decision tree is greatest frequently castoff method in predicting learner’s performance. Data mining in the learning field recognized as EDM (Educational Data Mining). It happens since the escalation in learning assets and records that can be discovered to study how a learner educated. Scientists also examined the features that impact education results [33] Table 6 shows the accuracy result for the DT (Decision Tree), GL (Generalized Linear) and DL (Deep Learning) with runtime.

Security

Exhausting unlicensed APs in the government, military services and business, there is a great opportunity of presence subject to different hacking and viruses occurrences. It requires to identify unofficial APs for the security of data. Innumerable approaches of investigation are in development, addressing several features of the problem [36]. Dataset was created using RTT (Round Trip Time) ethics. The facts set organized to apply ML algorithm to give the outcome, and then outcomes gained are matched, to display which procedure is more correct [35]. Table 7 shows the various ML methods are used for security purposes with its accuracy.

Current Scenario

A lot of research has been done exploring basic ML techniques. Research is now shifted to advanced algorithms. We now present the state-of-art of the latest trends in ML. Advanced topics in ML including Nonlinear Dimension Reduction, Meta-Learning, Linear Modelling, Feature Selection, Kernel Selection, Exponential Family Models, Multi-Task Learning, Graphical Models, Graph-Based Semi-Supervised Learning, Structured Support Vector Machines, Maximum Entropy, Conditional Random Fields, Approximate Inference, Boosting and Clustering.

Conclusion

This study shows that ml techniques are crucial in different fields. ML has provides different techniques and various methods for getting the correct result. It also confirms that ml methods are exclusively used in the diagnosis and predictions of various types of diesel. ML skills for estimation of several illnesses like heart, cancer, diabetic, thyroid disease, etc. using SVM, Nave Bayes and DT. For heart disease, we get 86.41% correct result by using the Nave Bayes method. In the finance sector ml provide tool “halo” by PricewaterhouseCoopers. For predicting student’s performance, implementing the generalized linear method is 66.6% right with run time 1s. Security purpose it needs to find unofficial Aps for the protection of data. C 4.5 gives 92.9% of total correctness. At last, discuss advanced machine learning classifiers to solve past problems and get more accurate results with less runtime.

Cite this essay

Machine Learning. (2019, Dec 17). Retrieved from https://studymoose.com/machine-learning-applications-techniques-and-current-example-essay

How to Avoid Plagiarism
  • Use multiple resourses when assembling your essay
  • Use Plagiarism Checker to double check your essay
  • Get help from professional writers when not sure you can do it yourself
  • Do not copy and paste free to download essays
Get plagiarism free essay

Not Finding What You Need?

Search for essay samples now

image

Your Answer is very helpful for Us
Thank you a lot!