0% found this document useful (0 votes)
10 views20 pages

SIT720 Week 1

The document provides an introductory overview of Machine Learning (ML), covering its real-world applications, definitions, types, and model assessment. It discusses various applications such as robotics, board games, voice recognition, and digit recognition, along with the steps and data representation involved in ML. Additionally, it explains the differences between unsupervised and supervised learning, and the importance of model evaluation and selection.

Uploaded by

syedsaudnaqvi
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
10 views20 pages

SIT720 Week 1

The document provides an introductory overview of Machine Learning (ML), covering its real-world applications, definitions, types, and model assessment. It discusses various applications such as robotics, board games, voice recognition, and digit recognition, along with the steps and data representation involved in ML. Additionally, it explains the differences between unsupervised and supervised learning, and the importance of model evaluation and selection.

Uploaded by

syedsaudnaqvi
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Foundation of ML

Week 1
List of contents
Introductory overview of:
 Real-world applications of Machine Learning (ML)

 Definitions of ML

 Types of ML

 Unsupervised learning

 Supervised learning

 Linear models

 Nonlinear models

 Model assessment and selection


Real-world applications of machine
learning
 ML involves “computer algorithms” that learn how to
perform different tasks
 Robotics
 Board Games
 Voice Recognition
 Digit Recognition
Examples - Robotics
 ML is a fundamental part of robotics for enabling robots
to perform -
 household work ranging from cleaning, cooking, reading and
scheduling tasks
 Simultaneous Localization and Mapping (SLAM)
 walking patterns of humanoid robots
 finding routes for rescue robots
Examples - Board Games
 one of the oldest applications of ML
 in March 2016, AlphaGo, the board-game-playing AI from
Google’s DeepMind played Korean Go Champion Lee
Sedol
 AlphaGo won the game
4 points to 1.
Examples - Voice Recognition
 benefited from advances in deep learning
 as well as big data
 Siri uses
 speech recognizer,
 natural language processing
 text-to-speech techniques
Examples - Digit Recognition
 The task of reading in the images of handwritten
numbers and letters
 Recognise the digits
 output the machine-encoded equivalent
 ML methods (SVM and Deep Learning) have hit >99%
accuracy for this task
Definitions of ML
 “Field of study that gives computers the ability to learn
without being explicitly programmed,” (Samuel 1959)
 If not explicitly programmed can it learn to do things?
How? Magic?
 “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,” (Mitchell 1997, p. 2)
Steps in ML
 How do we learn to perform a task?
 have access to data from which we can learn (Data
Manipulation)
 find patterns or build the model (Analytics)
 finally, evaluate the model and visualise results (Evaluation
and Visualisation)
Data representation
 Text data
 Without tools, it’s difficult for humans to analyse and interpret
larges volumes of data
 ML requires data to be described by attributes or parameters
prior to learning a model
Without tools, it’s difficult for humans to analyse
and interpret larges volumes of data.. Without tools,
it’s difficult for humans to analyse and interpret
larges volumes of data. ML requires data to be
• Number of vowels
described by attributes or parameters prior to
learning a model . Without tools, it’s difficult for
• Frequency of a given set of words
humans to analyse and interpret larges volumes of • Length of the document
data. ML requires data to be described by attributes
or parameters prior to learning a model. ML • Set of repetitive words
requires data to be described by attributes or
parameters prior to learning a model. Without tools, • Frequency of words
it’s difficult for humans to analyse and interpret
larges volumes of data. ML requires data to be
• Number of sentences
described by attributes or parameters prior to
learning a model. Without tools, it’s difficult for
• Number of adjectives
humans to analyse and interpret larges volumes of • Number of positive and negative words
data. ML requires data to be described by attributes
or parameters prior to learning a model.
Data representation...
 Image data
 build a system able to identify if a given image is from outdoors or
not
 needs to be represented in a vector of features
 consider a image divided into 9×15=135 blocks
 For these blocks we can compute – Mean, Variance, radiant,
other statistics
 p features per block leads to
 135p features per image
 for n images, the size of Feature Matrix is 135p×n
ML Type - Unsupervised learning
 How do you find the underlying structure of a dataset
which is unlabelled?
 Popular approaches - Clustering (similarity-based)
 the process of grouping similar points together
 gives insight into underlying patterns of different groups
ML Type - Unsupervised learning...
 Common examples -
 Data understanding and visualization
 Anomaly detection
 Information retrieval
 Data compression (reduction)
ML Type - Supervised learning
 “Learn a function (model) from data to relate the inputs with
outputs.”
 In supervised learning, the training data includes output
information (labels/targets)
 Target function: f:X→Y
 Examples: It is in the form of (x,y), denoted as (x1,y1), ...,
(xn,yn)
 Hypothesis g:X→Y such that g(x)=f(x)
 x = set of attribute values
 y = discrete label (classification), real valued number (regression)
ML Type - Supervised learning..
 Classification problem
 with two classes, decision boundaries are a hyper-surface that
partitions data space into two sets
 each of these sets represents one of the classes
 linear vs non-linear decision boundary
ML Type - Supervised learning...
 Regression problem
 to examine the relationship between response variables and
one or more predictor variables
 examination can result in a hyperplane, representing the
regression analysis
 regression problem in 2 dimensions
Model assessment and adjustment
 Model evaluation
 to determine if it will do a perfect job of predicting the labels on
new and future test data
 randomly split examples into a training set and test set
 use training set to learn a model
 evaluate the model using test set and a measurement (such as
accuracy of prediction)
 repeat for different random splits and
average results
 more training data, more accuracy
Model assessment and adjustment...
 Model selection
 how to find the BEST model (hypothesis)?
 There are often many knobs (parameters and hyper-
parameters) that we can use to vary its fitness to the data
 effective ways in which people approach this problem
 look at averaged evaluation score on many random test sets
 cross-validation (train using one set and test on the other, rotate
them) etc.
 be aware of *Over-fitting*
Summary
Data Model

Decision
Thank You.

You might also like