Machine Learning: Question Set – 03

What is Human learning?

In In cognitive science, learning is typically referred to as the process of gaining information through observation . It is the result of repeated exposure to new things or situations and is an active process often involving some degree of learning from the past.

A cognitive process, which is different from a conscious experience, can be induced by exposing one to a set of stimuli, usually during a learning task, and then evaluating how the information is processed, how it is remembered, or how an individual remembers more often in the latter case.

An individual can learn from past experiences by observing what has happened to the individual. This can take the form of a passive process in which the learner is passively receiving information

Discuss different ways of human learning

  1. Subject/domain expert teaches us directly
  2. Build our own notion directly from the knowledge we got in past
  3. Learn by ourselves

Parents shows the objects of different colors to kid and teach him what color or what shape it is. It comes under the direct teaching by expert. Or parents teach him alphabets by showing or writing A, B, C, etc.

When kid grows up and start attending school, he learns forming the words or sentences from learned alphabets. This is how he starts building his own vocabulary from past learning he got from his parents.

When he went to higher education, he starts taking examinations and he has to write the answers of question papers. The knowledge shared by teacher in classroom and the questions asked in examination may be quite different. Student has to made appropriate assumption and think about the possible answers. This is in fact self learning.

Ways of human learning
Ways of human learning

State the brief history of evolution of Machine Learing

Time line of evolution of machine learning is shown below:

  • 1950: Alan Turing in his seminal work entitled ‘Computing machinery and Intelligence’ (Mind, New Series, Vol. 59, No. 236, Oct, 1950, pp. 433-460) posed a question “Do machines have intelligences?”
  • 1952: Sir Arthur Samuel developed very first machine learning program which could play the Checkers game
  • 1957: Frank Rosenblatt designed very first neural network program which mimics the functionality of human brain at very primary level.
  • 1967: Start of pattern recognition – Invented Nearest neighbor algorithm
  • 1979: First self driving car was developed at Sandford University which can navigate and avoid obstacles in room
  • 1982: Recurrence Neural Network came into existence
  • 1989: Conceptualization of Reinforcement learning. Era of commercialization of Machine Learning started
  • 1995: Two very popular algorithms were developed: Random Forest and Support Vector Machine
  • 1997: IBM’s Deep Blue beats the world chess champion Gary Kasparov
  • 2006: Netflix launched a first machine learning competition. Deep learning is conceptualized by Geoffrey Hinton
  • 2010: Machine learning website – Kaggle was launched
  • 2011: IBM’ Watson beats two human champion in Jeopardy
  • 2016: Google’s AlphaGo beats the unhandicapped professional human player
History of Machine learning and Human Learning
History of Machine Learning

Image Source: https://www.zeolearn.com/magazine/what-is-machine-learning

State the formal definition of Machine Learning

Machine learning can be formally defined as,

A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E

– Tom M. Mitchell

Discuss the steps involved in Machine Learning

We can summarize machine learning process by three steps.

  1. Input: Historical data is used to train the model
  2. Abstraction: Raw data is represented in broader way using ML algorithms
  3. Generalization: Trained model is generalized to make decisions. Model then accurately predicts the results for unseen data

Let us understand machine learning process from human perspective. Machine should be able to infer the rules from the presented input data. At broader level, it should be able to perform the desire operation for any data.

When student memorize something, he can easily answer the questions asked in examination from that list. He will simply write the things he has memorized before.

Additional Reading: What makes human learning different from machine learning? Click to read.


≪ Previous Page | Next Page ≫