Machine Learning (Self- Study) — 1

A Basic Overview & stuffs ;)

Before starting to talk about what is M.L , it’s applications & all that stuffs, Let’s talk about the Pre-requisite of a Machine Learning Journey. Unlike most other IT Jobs a M.L Engineer not only have to have the coding skills but also a good grasp in Mathematics. Without mathematics studying M.L is like sabotaging a plan before even starting it. But on the bright side to start with our M.L journey we don’t need to be a huge mathematician or something like that some basic understandings will be enough to get started. So, let’s see what the pre-requisites are :

1. Basic Statistics
2. Basic Calculus
3. Basic Linear Algebra
4. Basic Python

Let’s check if we got our basics clear or not.

1. Matrix Multiplication ->

Multiply the Metrcies

2. Calculate Derivative ->

Calculate the Derivative

3. Count the numbers of letter, word :

Write a python program to count letters and words from a sentence using function (avoid using any pre defined functions)

Write The function

Solution :

Matrix Multiplication Solution

Derivative Solution

Count Words, letters Solution

What is Machine Learning ?

Machine Learning is described as many lot of things like next natural step of Artificial Intelligence, application of Statistics, some advance algorithms etc. But in reality M.L is all of them . Like A.I ML is about understanding, processing data and replying intelligently based on that , like stat it evaluates data using various advanced formulae, like algo it provides o/p based on some i/p data .

Machine Learning v/s Traditional Programming

Let’s take an example,

In traditional programming to distinguish something or some one let’s say to recognize me through traditional programming the program need to look for glasses, black hair & other distinguishable stuffs which are need to be mentioned properly (hardcoded in the program).

whereas in M.L all we gotta do is pass numerous images of our face and automatically it starts to recognise me through various algo .

Why M.L is now feasible now but wasn’t a decade back?

To work on M.L models we need lots of data and to work with that we need high computational ability. But if we go back a decade or two we hardly find any proper amount of data that can be used and also the computational power back in the days were much lower compared to now.

But now almost everything is digitalised making a huge amount of data that can be used & with the upgrades in computational power it’s much easier to work with M.L nowadays.

Application of Machine Learning :

Use of Machine Learning can be seen everywhere from sports, health care, education, stock market and many more . Actually it’s really hard to find any field where there isn’t any touch of M.L.

Machine learning mainly have two aspects :

1. Artificial Intelligence
2. Data Science

What is Artificial Intelligence? Where did ML grew apart from AI?

Artificial Intelligence is usually meant by a non human agent(computers) that can mimic human ways of thinking and doing things in real world. But the main difference between AI and ML is instead of acting as humans , ML is of broader sense which can be described as doing autonomous things which may or may not be same as like humans.

Fundamental Conundrums of Artificial Intelligence :

1. Intelligent agents have limited resources.
2. Computation is local, but problems have global constraints.
3. Logic is deductive, but many problems are not.
4. problem solving, reasoning, and learning are complex, but explanation and justification are even more complex.

Fundamental Characteristics of AI Problems :

1. Knowledge often arrives incrementally.
2. Problems exhibit recurring patterns.
3. Problems have multiple level of granularity.
4. Many problems are computationally intractable.
5. The world is dynamic, but the Knowledge of the world is static.
6. The world is open ended but the knowledge is limited.

AI as UNCERTAINTY Management

AI = What to do when you don’t know what to do?

Reasons of Uncertainty :

1. Sensor limits
2. Adversaries
3. Stochastic environment
4. Laziness
5. Ignorance

Example Problems of AI :

1. Answering Jeopardy questions.
2. Configuring dimensions for the basement of a new house.
3. Tying Shoelaces
4. Deciding on a route to a new destination.
5. Making sense of a news broadcast.
6. Designing robots that walks on water ! ;)
7. Establishing a flower pot that can be used as a drinking cup ;)
8. Deciding whether or not a new animal is a bird.

Before moving on we must first know about Bayes rule. Many of us already know what it is. So, I’m gonna briefly touch this topic. OK! so the first question arises is why I’m talking about Bayes rule all of a sudden?. So for starters Bayes theorem is one of the foundational element of Artificial Intelligence and a critical concept of ML as well. Before we start understanding Bayes theorem needs some basic statistics understandings which I think is we all know .

Bayes Rule :

where A & B are events . And P(B) ≠ 0

• P(A | B) is a conditional probability: the probability of event A occuring given that B is true. It is also called the Posterior Probability of A given by B.
• P(B | A) is also a conditional probability : the probability of event B occurring given that A is true because P(B | A) = L(A |B)
• P(A) & P(B) are the probabilities of observing A and B respectively without any given condition; they are known as marginal probability or prior probability.
• A and B must be different events.

What is Data Science ?

Data science is the domain of study that deals with vast volumes of data using modern tools and techniques to find unseen patterns, derive meaningful information, and make business decisions. Data science uses complex machine learning algorithms to build predictive models.

Who is a data Scientist?

Basic Data Scientist Skills :

Substantive expertise :

— Knows which questions to ask

— can interpret the data well

— understands structure of data

— Data Scientists often works in teams

Problems solved by Data Science:

1. Movie recommendation (in sites like Netflix, Amazon prime)
2. Social Media (Facebook, Instagram and alot more)
3. Web Apps ( Tinder, Uber etc.)
4. Bio informatics (Understanding gene data)
5. Urban Plannings
6. Astrophysicists (Analysing start clusters and a lot more)
7. Public Health
8. Sports

Induction & Deduction :

Induction -> Given a cause & effect, induce a rule.

Deduction -> Given the rule and the cause deduce the effect.

Abduction -> Given a rule and effect, abduce a cause.

Machine Learning is basically Induction because in ML given the cause, effect data we try to induce a proper relation/rule which is used to predict unknown effect for different cause.

What is Machine Learning ?

The use and development of computer systems that are able to learn and adapt without explicit instructions given to them, by using algorithms and statistical models to analyse and draw inferences from patterns in data.

Types of Machine Learning :

There are mainly 3 types of ML algorithms->

1. Supervised Machine Learning
2. Unsupervised Machine Learning
3. Re-inforcement Learning

Types of ML

Supervised Learning : Supervised learning is the types of machine learning in which machines are trained using well “labelled” training data, and on basis of that data, machines predict the output.

Unsupervised Learning : Unsupervised learning is a type of machine learning in which models are trained using unlabelled dataset and are allowed to act on that data without any supervision

Reinforcement Learning : Reinforcement learning works on a feedback-based process, in which an AI agent (A software component) automatically explore its surrounding by hitting & trail, taking action, learning from experiences, and improving its performance.

In this blog I tried to give a basic overview of Machine Learning. In the next blog I will explore the Titanic Survival Exploration & after that I will get deeper into the theories & practicals.