Let's Kickstart
ML
Nishan Pantha / @nishparadox
GitHub/ NISH1001
nishanpantha.com.np

What This Talk About

What ML hype is about?

Guide for beginners

Methods used

Okay! I want to get involved. But how?

General problems we face while doing it

But first...

what is Artificial Intelligence?

Building systems that can do intelligent things

Intelligence is

Perpective

Psychological

Ambiguous

Map

Translate

Alpha Go

Youtube's Recommendation

Facebook's News Feed

Other

  • Spam Filter in Email
  • Google Search
  • Weather Prediction
  • Games

What is Machine Learning

one of the mechanisms to AI

Building systems that learn from experience

Different methods used to LEARN

Types of Learning

Supervised Learning

- Train the system

- Student learning from a teacher

Training Dataset

- n attributes

- k classes/target

Glucose BP BMI Age Diabetes(Y/N)?
148 72 33.6 50 Y
85 66 26.6 31 N
97 66 23.6 22 N

...

...

Cat vs Dog

OCR

Supervised Learning

Unsupervised Learning

- Let the system figure out the patterns itself

- Students learning on their own

Dataset

- n attributes

- no target class/label

Usage

- Topic modelling

- Unusal activity detection

- Grouping similar texts

Reinforcement Learning

- Rewards based learning

I don't know how to act in this environment, can you find a good behavior and meanwhile I'll give you feedback.

Categories based on Outputs

Classification

Outputs as class labels

Hot Dog or Not?

Regression

Outputs as continuous values

Find relationship among variables

Y = F(X) ?

Linear Regression

Clustering

Unsupervised

Data Intensive

Data?

Computation Intensive

How do I start ?

Base Knowledge

Linear Algebra

Probability and Statistics

Multivariate Calculus

Programming

Data Structures and Algorithms

Linear Algebra

- Matrices

- Vectors

Probability and Statistics

- The Basics

- Bayes Theorem?

- Distribution

- Hypothesis Testing

Take an online course

  • Machine Learning by Andrew NG
  • Linear Algebra By Gilbert Strang
  • Neural Networks for Machine Learning By Geoffrey Hinton

There's more

Bottom Up Learning

Top Down Learning

Implement. Implement. Implement.

Don't Hesitate To Code

Keep on Experimenting

Learn. Share.