Challenges before existence of Machine Learning:

Before machine learning became so widely used AI systems were mostly rule based which  means that these rules failed in real life scenarios, most of these state representations are been manually coded  which is hand coded as everyone know hand coding is quite difficult and the main disadvantage was that these problem scenarios or test scenarios would fail outside for  what the rules were coded for that means they would not be useful in real life application even though work well  in their controlled environment.

Why Machine Learning:

Machine Learning has changed our lives in several significant ways, it allows powerful processing which means it can process  through  far more complicated data hence decision making awaiting will much more well founded and the predictions are much more accurate, it not only allows powerful processing but also quicker processing that is more work is done in quicker time. When we think about storage of large sum of data, with machine learning  we can perform  affordable data management . Machine learning method is also considerably less expensive. Finally the important thing is when we get more and more data the more and more complexity it becomes, machine learning allows us to analyze these complex big data.

What is Machine Learning:

Machine learning is a method that allows computers to imitate and adapt to human life behavior, so the machine analyzes the past data learns from the data and make decisions or predictions so the computer  is creating  its own logical solutions all of which without any human assistance, these systems will grow, develop themselves when exposed to new data.

Categories of Machine Learning Algorithms:

This algorithm stands on two primary pillars known as supervised learning  and unsupervised learning. Some humans also remember a new discipline of study—deep learning—to be break away the question of supervised vs. Unsupervised learning.

Supervised learning is when a computer is offered with examples of inputs and their favored outputs, the purpose of the laptop is to be taught a common formulation which maps inputs to outputs. This can also be further broken down into:

Semi-supervised studying : which is when the computer is given an incomplete training set with some outputs missing

Active learning: which is when the laptop can most effectively receive coaching labels for a very limited set of instances? When used interactively, their training sets will also be awarded to the consumer for labeling.

Reinforcement learning : which is when the learning information is simplest given as feedback to the software’s actions within the dynamic environment, equivalent to using a auto or playing a sport in opposition to an opponent

In contrast, unsupervised studying is when no labels are given at all and it’s as much as the algorithm to search out the structure at it’s enter. Unsupervised finding out generally is a purpose in itself after we best ought to detect hidden patterns. Deep studying is a brand new field of study which is encouraged by using the structure and performance of the human mind and based on artificial neural networks as an alternative than just statistical principles. Deep finding out can be used in both supervised and unsupervised strategies.

Future of Machine Learning:

Deeper Personalization: It means Direct marketing towards your interests and advertisements based on your profiles.

Self Driving Cars : Although it’s the concept that’s been worked out  right now it’s not one that reaches completion but in the near future we might have cars that can run flawlessly even on the most crowded roads.

Smarter Investment Opportunities : Machine learning could ensure that your profit is maximized taking into consideration of your past purchases and the current market scenario to suggest smarter investment opportunities.

Machine learning provides greater opportunities in terms of the job that it can provide. Google trends shows increased interest in machine learning, machine learning market size is also expected to grow through 1.03 billion dollars to 8.81 billion dollars by the year 2022; this indicates a bright and clear future for individual skill in the concept of Machine learning.

Course duration : 120 hours (Only on Saturday & Sundays)

Course Fee : 40,000/-

Pre-Requisites (Eligibility):

  1. Programming :
    1. Data Structures: Arrays, Trees
    2. OOPs(Object Orienting Programming Principles)
  2. Python(Basics) :
    1. Looping
    2. conditionals (If else)
    3. Functions (Writing functions and calling them)
    4. Class
    5. Data Structures: lists, tuples
  3. Mathematics:
    1. Probabilities
    2. Statistics (good grip on statistical concepts- class 12 level, probability distributions)
    3. Sets
    4. Linear programming

Preferred (not necessary):

  1. Versioning :
    1. Git
  2. Python:
    1. Libraries- numpy, pandas
  3. Mathematics:
    1. Statistical distribution and their applications – engineering mathematics
    2. Calculus
    3. Linear Algebra
  4. Programming:
    1. Fuzzy theory and Fuzzy networks

Course Structure:

  1. Introduction to Machine Learning mathematics – Brief Revision of important Mathematics concepts (2 Hours)
  2. Introduction to Machine Learning – What is ML and The need of ML (1 Hour)
  3. Major terminologies in Machine Learning. (2 Hours)
  4. Templating any machine Learning Project (1 Hour)
  5. Explaining setting up Dev environment – advantages of using a managed environment and network shareable notebooks (2 Hours)
  6. Setting up github accounts and Introduction to Git— necessary for: (3.5 Hours)
    1. Internal Evaluations
    2. Allows students to provide direct project links in resumes
  7. Overview of Python in Machine Learning – View of the Python Communities, Checking New Projects. Understanding the power of github for self learning. (1 Hour)
  8. Hands on experience with Numpy and Pandas – Integral libraries in Machine learning (3 Hours)
  9. Hands on with a Dataset (testing our setup and initializing weekly code assignment submissions ) basic statistical operations on the data set (2 Hrs)
  10. Introduction to Fuzzy theory (3 Hours)
  11. Concepts of Data handling:
    1. Data Cleaning (4.5 Hours)
    2. Feature Selection (3 Hours)
    3. Feature Engineering (2 Hours)
  12. Introduction to types of learning (1 Hour)
    1. Supervised
    2. Unsupervised
    3. Reinforced
  13. Supervised Learning
    1. Naive Bayes (2 Hours)
    2. Decision Trees (2 Hours)
  14. Introduction to regression –theory (1 Hour)
  15. Regression Models: (3 Hours)
    1. Linear
    2. Polynomial
    3. Logistic
  16. SGD – Stochastic Gradient Descent (3.5 Hours)
  17. Support Vector Machines (4 Hours)
    1. Kernels
    2. Implementation
  18. Unsupervised learning
    1. K-MEAN (2 Hours)
    2. Mean Shifting ( 2 Hours)
    3. Birch (2 Hours)
  19. Ensemble methods (3.5 Hours)
    1. Boosting
    2. Bagging
  20. Cross validation (2 Hours)
  21. Parameter Tuning
    1. Grid Search (3 Hours)
    2. Random Search (3 Hours)
  22. Deploying ML models in real world (3 Hours)

Key features of the training:

  1. Github based submissions will allow students to directly showcase work to recruiters.
  2. Bi monthly 1 hr discussion/ presentation sessions where students can discuss and present topics outside the course structure to enhance learning.
  3. Weekly coding assignments (evaluation is based on effort rather than accuracy scores)
  4. 1 capstone project – one complete ML project that students can implements from scratch (teacher will guide them to select appropriate Dataset)