Wednesday, December 27, 2017

The Master Algorithm - Pedro Domingos

A description of the state of present day machine learning organized by the main techniques/algorithms, the domains from which the techniques originated, the history of the field and how the search for the master algorithm continues.

1. The Master algorithm

  • The AI algorithm that can replicate/reproduce all of human knowledge.
  • Areas that might yield the master algorithm: NeuroScience, Statistics, Evolution, Physics, Computer Science
  • 5 approaches to AI (and the fields that influenced them): Bayesian (Statistics), Symbolist (Inverse deduction) , SVM (Analogies), Connectionist (Neuroscience), Genetic programming/Evolutionary algorithms (Evolution)

2. The Humes Inference question

  • Can everything be inferred from limited knowledge? Can the past ever be used to accurately predict the future?
  • Is a Master Algorithm feasible?
  • Rationalism (All knowledge comes from reasoning)  vs empiricism (All knowledge comes from experimentation/observation)
  •  The Symbolist approach:
    •     Knowledge is just the manipulation of symbols
    •     Inverse deduction: Deduce complex rules from simple rules
    •     E.g. Decision tress

3. The Connectist approach

  • Neural networks:
    • Theory behind how the brain learns (Hebbs): Axions fire across neurons, reinforced every time they are fired, developing a memory
  • Perceptron: Weighted inputs + Thresholding function
    • Drawback: Can classify only when there is a linear boundary
    • E.g. XOR has three regions 00, 01, 10, 00
    • E.g. Gender + Age (0/1): How to classify if condition is true for Male/Young and Female/Old but not for other cases
  •  Neural networks: Multilayer perceptrons with backprop to learn
  •  Others: Autoencoders, Boltzmann machines, CNNs

4. The Evolutionary/Genetic approach

  • Based on genetic algorithms
  • A set of solutions, combined continuously in iterations, at each iteration, weakest solutions are discarded
  • Iteration continue until an optimal solution is reached.

5. The Bayesian approach

  • Based on Bayes' rule: Probability of cause of an effect, given probability of cause, probaility of effect, and probability of effect given cause.
  • P(cause|effect) = P(cause)*P(effect|cause)/P(effect)
    •    P(cause): Prior
    •    P(cause| effect): Posterior
    •    P(effect | cause)/P(effect): Support
    •    Imagine a Venn diagram: 4 areas: C, E, C and E, Not C/Not E => Four probs: E, C, E|C, C|E
  • Bayesian networks: Networks of Bayesian infererers: Used to infer probabilites of complex sequences
  • E.g. Hidden Markov Models, Kalman Filters (Continuous variable version of discrete HMMs)

6. The Analogist approach

  • Find similarities between examples
  • E.g.Nearest Neighbor, SVM

7. Unsupervised approaches: Learning without teaching

  • E.g. KMeans, Principal Component Analysis, Reinforcement learning, Relational learning

8. Unifying the algorithms

  • Metalearning: Techniques to combine approaches. All reduce variance
    • Random forests:Bootstrapping, Bagging, Boosting

Thursday, October 26, 2017

Little Bets - Peter Sims

Little bets:

  • Small decisions/goals/tasks
  • Large companies fail because they look for large billion dollar markets rather than experimenting with smaller markets that might grow: E.g: HP
  • Affordable loss principle: Make decisions based on  what you can afford to lose, rather than on expected gain

Growth mind set:

  • Large number of small attempts with multiple failures, rather than one large bet
  • Fixed mind set: Abilities/intelligence is fixed. Growth mind set: Results are determined by effort, not intelligence. 

Fail fast to learn fast:;

  • Healthy perfectionism: Driven by desire for excellence. Unhealthy perfectionism: Driven by fear of failure
  • Fail fast through little bets around prototypes

Genius of play:

  • Environments that lead to improvisation result in creativity
  • Plussing (Pixar): Idea evolution in a team is through a series of "ands" rather than "buts"

Problems are the new solutions:

  • Break large problem into smaller problems. E.g. Walt Disney Concert Hall, Agile development, McMaster's Iraq strategy

Questions are the new answers

  • Need to go out into the world and ask questions to find the problems. E.g. Grameen Bank, McMaster's Iraq strategies
  • Encourage voracious questioning

Learning a little from a lot

  • Learn from everyone, to get different persepctive
  • Build an open network of diverse people, and maintain it to constantly receive different perspectives

Learning a lot from a little

  • Seek out active users (early adopters). They provide 75% of improvements you will need

Small wins

  • Little bets can lead to small wins. Small wins can lead to successes.

Sunday, January 22, 2017

Confucius in 90 minutes - Paul Strathern

Confucius (Kungfutzu)

  • Circa 600BC in North central coastal China. 
  • Started a successful school for bureaucrats
    • Taught his philosophies of conduct and ethics. 
    • Students were often sons of rulers. 
  • Later in his life traveled through China,  meeting and advising rulers of various states.


  • A philosophy that evolved out of his teachings. 
  • Teachings were practical rather than religious or metaphysical. 
  • Dealt with the conduct and morality of rulers, bureaucrats, and citizens . 
  • Central premise: Ordinary activities of individuals are sacred and must be conducted in an ethical manner. 
  • Central concept: "jen" - a quality of magnanimity, virtue and honesty which every individual should strive for. 
  • The goal was to produce a society of individuals who live a life of harmony and virtue.
  • Contrast with the other major philosophy of the time, Taoism ("the way"), which dealt with metaphysics.
  • Encapsulated in pithy sayings documented in his books of sayings - the Analects. 
  • Confucianism has other sacred texts (some predated Confucius, others were edited by his followers):
    • Four books (one of which is the Analects)
    • Five Classics (among which are IChing: Book of Changes, dealing with metaphysics and the cosmos as an interaction of yin and yang, the Book of Poetry, the Book of History).