Machine learning: A strategy to learn and understand


Simple, English explanations to learn and understand the concepts of machine learning and its applications.

Road-map

An easy and simple way of learning and understanding machine learning. Part 1: Why Machine Learning Matters. The big picture of artificial intelligence and machine learning — past, present, and future.

Part 2: Supervised Learning.

Part 3: Unsupervised Learning.

Part 4: Neural Networks & Deep Learning. Why, where, and how deep learning works. Drawing inspiration from the brain. Convolution neural networks (CNNs), recurrent neural networks (RNNs).

Part 5: Reinforcement Learning.

Appendix: The Best Machine Learning Resources. A curated list of resources for creating your machine learning curriculum.

Why machine learning matters?

Artificial Intelligence is shaping our futures at an astonishing speed. And it will change the world completely than any other innovation or discovery in this 21st Century. Anyone who doesn’t know and understand it will soon find themselves left behind, waking up in world full of futuristic technology which is currently feels like a magic or a science fiction.

After couple of ups and downs in the field of AI, the rapid development in the data storage technologies and computer processing power has dramatically change the game in recent years.

Google trained its controversial AI agent in 2015, which not only can talk to a person but can also discuss morality, express opinions and answer general facts based questions.

And in the same year 2015, the DeepMind developed an AI agent that surpassed the human level performance at 49 Atari games. Soon after in 2016, DeepMind obsoleted its own previous record by releasing a new state of the art game play method called A3C. Meanwhile DeepMind’s AlphaGo defeated the world GO champion. That was an extraordinary achievement in a game which was dominated by the human players after machines first conquered Chess. Even now World renowned masters of GO game not understood that how it is possible for a machine to fully understand every possible moves of the game, which is the most ancient Chinese war strategy game. Which has a possibility of 10¹⁷⁰ board positions.

Professional Go player Lee Sedol reviewing his match with AlphaGo after defeat.

In March 2017 OpenAI created its AI agents that invented its own language. Soon after Facebook reportedly trained its own AI agent which can negotiate and even can lie. And on August 11, 2017 OpenAI achieved another milestone by defeating human champions of online multi-player game DOTA2.

Dota2 match between a human vs OpenAI (bot).

Today AI is used to discover treatments for cancer patients, instantly analyze results from the medical reports and conduct scientific research for drug discovery. AI also has its involvements now even in the smart phone applications like AI beautification modes in a camera application. Companies even built the dedicated AI chip for smart phones. i.e. Qualcomm, Huawei, Apple etc.

In everyday life AI is taking roles which was previously occupied by humans. And it is with no doubt doing it better than humans with more efficiency and accuracy.

Artificial Intelligence vs Machine Learning

Artificial Intelligence is the simulation of human intelligence processes by machines or computer systems. Its foundation includes mathematics, statistics and probability, logic and philosophy, linguistics, neuroscience and decision theory. AI also has other sub-fields too like robotics, machine learning, deep learning, computer vision and natural language processing.

Machine learning and Deep learning are the sub fields of AI.

Machine learning deals with the algorithms which have the ability to learn by itself without explicit programming rules and models. A machine learning algorithms learns through the given models and try to predict things.

Deep Learning is the sub field of AI and subset of machine learning in which artificial neural networks or ANNs adapts and learn from big data and models.

A recent report by the Future of Humanity Institute surveyed a panel of AI researchers on time-lines for AGI, and found that “researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years”.

Advent of super AI which would surpasses the humans in near future for good or worse to our specie. It carries with it the immense challenge of specifying what AIs will want in a way that is friendly to humans.

While it is difficult and early to say what future holds, but one thing is clear that it is the good time to start learning and understanding AI and machine learning. Before AI make us slave (Just kidding). Machine learning is at the core of our journey towards artificial general intelligence, and in the meantime, it will change every industry and have a massive impact on our day-to-day lives.


More on The 21st Century’s Machine learning: A strategy to learn and understand.

Part 2: Supervised Learning.

Part 3: Unsupervised Learning.

Part 4: Neural Networks & Deep Learning.

Part 5: Reinforcement Learning.

Appendix: The Best Machine Learning Resources.

(Will be updated soon on The 21st Century) Stay connected.

Support us on Patreon so we can grow more!

Follow The 21st Century to get the further readings on Machine learning: A strategy to learn and understand.

Source: Deep Learning on Medium