the essentials of machine learning

13
As technology marches on with its conquest of miracles we find Machine Learning becoming more and more ubiquitous. From smartphones to chatbots (remember the recent controversy surround an AI chatbot of Microsoft) machine learning is fast Machine Learning

Upload: dexlab-analytics

Post on 12-Apr-2017

249 views

Category:

Career


2 download

TRANSCRIPT

Page 1: The Essentials of Machine Learning

As technology marches on with its conquest of miracles we find Machine Learning becoming more and more ubiquitous. From smartphones to chatbots (remember the recent controversy surround an AI chatbot of Microsoft) machine learning is fast becoming a part and parcel of our everyday lives in which technology plays a pivotal role.

Machine Learning

Page 2: The Essentials of Machine Learning

What is Machine Learning?

Machine Learning may present itself in the humblest of fashions like when cameras of smart phones are able to recognize faces of people. There are even simpler examples of our day to day interactions with machine learning. Suppose you have added the names and phone numbers of friends and acquaintances in your. And what happens when you start to dial a number the suggested contacts are displayed automatically. Here you unknowingly are teaching the phone to detect keywords and patterns.

Page 3: The Essentials of Machine Learning

Machine Learning Algorithms and Their Types

• There are number of ways in which Machine learning may take place and these are known as Machine Learning Algorithms.

• Three many types of Machine Learning Algorithms that dominate this evolving field. They are:

• Supervised Learning• Semi-supervised Learning• Unsupervised Learning

Page 4: The Essentials of Machine Learning

Important Keywords in Machine Learning

• However, before we move on with details of machine learning algorithms there are a number of keywords which we should be well aware of. These are:

• Training Data• Bayes Theorem and• K-Means

Page 5: The Essentials of Machine Learning

Training Data

Training Data: The data made available to the machine through input is known as Training Data as this data is used by the machine to further develop patterns. It consists of known labels technically called categorical variables like, ratings, gender and the like.

Page 6: The Essentials of Machine Learning

Bayes Theorem

Bayes Theorem: According to the Bayes Theorem, the product of probability of occurrence of event B and occurrence of event A when B has already occurred is equal to product of event A and occurrence of B when A has already occurred

Page 7: The Essentials of Machine Learning

K-Means

• Through the method of K-Means clustering based on Euclidean distance which is

• Here K is the no. of clusters.

Page 8: The Essentials of Machine Learning

Supervised Learning

Preparation of models are done through a process of training where the machine makes predictions and are corrected when they err. This process of training goes on till desirable levels of accuracy are achieved on the training data. Labels in training data are also present.

Page 9: The Essentials of Machine Learning

Example of Supervised Learning

To cite an example, past GRE scores and GPA of students in indicate that a score of 720 and a GPA of 4.2 will help them secure admission to good colleges. Inputting scores result in you being given feedback regarding whether you are rejected or selected. As abnormalities are present, this learning process is continuous. They also make use of Logistic Regression.

Page 10: The Essentials of Machine Learning

Unsupervised Learning

In case of unsupervised learning there is an absence of input data and the results are known beforehand. The preparation of the model occurs through deducing structures that are present in the input data. One of the goals may be to chance upon general rules. This process may occur in a mathematical manner so that redundancy is reduced or data is organized by similarity.

Page 11: The Essentials of Machine Learning

Example of Unsupervised Learning

To explain Unsupervised Learning we may cite the following example-You are engaged in cluster analysis in order to figure out the particular data points that form part of particular clusters. When a new data point is introduced the machines deduces it to be part of one of the clusters.

Page 12: The Essentials of Machine Learning

Semi-Supervised Learning

Here the input data is a blend of examples that may or may not have labels. A desired prediction problem is present but the model needs to learn the structures required for organizing data in addition to making predictions.

Page 13: The Essentials of Machine Learning

Thank You

DexLab Analytics would like to thank the viewers of this presentation for going through the same.

For details visit:

http://www.dexlabanalytics.com