Scribd Logo
Upload

Welcome to Scribd!

  • 12 views

    1 ModuleEcontent - Session5

    Uploaded by

    devesh verma
    01. Fitting functions to data points involves learning the relationship between features and targets using supervised machine learning techniques like regression and classification. 02. Classification involves using an object's characteristics to identify which class or category it belongs to, such as identifying emails as spam or not spam. Common classification techniques include binary classification, density estimation, and clustering. 03. Regression predicts a continuous valued output based on the values of input features. Polynomial curve fitting is a regression technique where the target is fitted with a polynomial function of varying degrees.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    1 ModuleEcontent - Session5

    Uploaded by

    devesh verma
    12 views24 pages
    01. Fitting functions to data points involves learning the relationship between features and targets using supervised machine learning techniques like regression and classification. 02. Classification involves using an object's characteristics to identify which class or category it belongs to, such as identifying emails as spam or not spam. Common classification techniques include binary classification, density estimation, and clustering. 03. Regression predicts a continuous valued output based on the values of input features. Polynomial curve fitting is a regression technique where the target is fitted with a polynomial function of varying degrees.

    Original Title

    1 ModuleEcontent_session5

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    01. Fitting functions to data points involves learning the relationship between features and targets using supervised machine learning techniques like regression and classification. 02. Classification involves using an object's characteristics to identify which class or category it belongs to, such as identifying emails as spam or not spam. Common classification techniques include binary classification, density estimation, and clustering. 03. Regression predicts a continuous valued output based on the values of input features. Polynomial curve fitting is a regression technique where the target is fitted with a polynomial function of varying degrees.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    12 views24 pages

    1 ModuleEcontent - Session5

    Uploaded by

    devesh verma
    01. Fitting functions to data points involves learning the relationship between features and targets using supervised machine learning techniques like regression and classification. 02. Classification involves using an object's characteristics to identify which class or category it belongs to, such as identifying emails as spam or not spam. Common classification techniques include binary classification, density estimation, and clustering. 03. Regression predicts a continuous valued output based on the values of input features. Polynomial curve fitting is a regression technique where the target is fitted with a polynomial function of varying degrees.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 24

    Outline

    01 Fitting Function to Data Point


    02 Classification

    03 Regression
    Fitting Function to Data Point

    Y= F(x)

    X Model y

    F(x)

    Given a data set D = {(xi, yi)} the objective is to learn the relationship
    between features and the target.
    Its a supervised machine learning technique.

    Use an object characteristics to identify which


    class/category it belongs to

    Classification Example:

    A new email is ‘spam’ or ‘non-spam’


    A patient diagnosed with a disease or not
    A short circuit fault had occurred or not
    Binary
    Classification
    ?
    𝒙
    Binary
    Classification
    𝐶𝐵
    ?
    𝒙
    Binary
    Classification
    𝐶𝑅
    𝐶𝐵
    ?
    𝒙

    Binary
    Classification
    𝐶𝑅
    𝑩 𝑹
    𝐶𝐵
    ?
    d(x, 𝑪𝑩 ) d(x, 𝑪𝑹 )
    𝒙

    Binary
    Classification d(x, 𝑪𝑩 ) > d(x, 𝑪𝑹 ) => 𝑹
    𝐶𝑅
    d(x, 𝑪𝑩 ) < d(x, 𝑪𝑹 ) => 𝑩
    𝑩 𝑹
    d(x, 𝑪𝑩 ) d(x, 𝑪𝑹 )
    --------- ---------
    𝐶𝐵 ? 𝑥 σ σ
    Binary
    Classification σ 𝒌
    (𝑿 𝑹
    𝒊 − µ 𝑹 )𝟐
    σ𝟐 = 𝒊=𝟏
    𝐶𝑅 𝒌
    σ𝒌𝒊=𝟏 𝑿𝑹
    𝒊
    µ=
    𝒌

    Density of the two classes don’t match


    Clustering
    𝑯𝒂𝒓𝒅 𝑪𝒍𝒂𝒔𝒔𝒊𝒇𝒊𝒄𝒂𝒕𝒊𝒐𝒏
    ?
    Hard and Soft = 𝒙

    Classification 𝑺𝒐𝒇𝒕 𝑪𝒍𝒂𝒔𝒔𝒊𝒇𝒊𝒄𝒂𝒕𝒊𝒐𝒏

    P( = ) = 0.6
    P( = ) = 0.4
    When h=0 ; P= 1/ (1+exp^(-h)) =1/2 =0.5
    h is large +ve value exp^(-h) is very small = 1/1= thus probability of getting a Blue is
    100%
    h is large -ve value exp^(-h) is very large thus probability =0 for getting a Blue is 0%

    𝑆𝑜𝑓𝑡 𝐶𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛
    ?
    𝒙
    1
    P( = ) =
    1+exp(−ℎ)

    ℎ = 𝑋2 - m𝑋1 - C
    Neural Networks

    ℎ = 𝑤1 𝑥1 + 𝑤2 𝑥2 + 𝑐
    v=𝜎 ℎ

    x1 ℎ′ = 𝑤′1 𝑥1 + 𝑤′2 𝑥2 + 𝑐’
    ∗ ∗
    ℎ∗ = 𝑤1 𝑣 + 𝑤2 𝑣 ′ + … . . +c
    ∗

    ∗ ∗
    v′ = 𝜎 ℎ′ v =𝜎 ℎ

    x2

    ℎ𝑚 = 𝑤𝑚1 𝑥1 + 𝑤𝑚2 𝑥2 + 𝑐m
    v𝑚 = 𝜎 ℎ𝑚
    Regression
    Polynomial Curve Fitting
    Polynomial Curve Fitting
    Polynomial Curve Fitting

    + 𝒘𝟑 𝒙𝟑
    Polynomial Curve Fitting
    Polynomial Curve Fitting

    Degree Training Data Test Point


    (M) Error Error
    0 0.9
    1 0.7
    3 0.1
    9 0
    Polynomial Curve Fitting

    Degree Training Data Test Point


    (M) Error
    0 0.9 0.2
    1 0.7 0.1
    3 0.1 0
    9 0 0.3

    Over Fitting
    References

    [1] STTP on ‘Basics of IoT and AI’ course Material, IIT, Kanpur

    [2] Christopher M Bishop: Pattern Recognition & Machine Leaning, Springer, pp. 362,
    2006

    [3]http://playground.tensorflow.org/#activation=tanh&batchSize=10&dataset=circle&reg
    Dataset=regplane&learningRate=0.03&regularizationRate=0&noise=0&networkShape=4,
    2&seed=0.58088&showTestData=false&discretize=false&percTrainData=50&x=true&y=tr
    ue&xTimesY=false&xSquared=false&ySquared=false&cosX=false&sinX=false&cosY=fal
    se&sinY=false&collectStats=false&problem=classification&initZero=false&hideText=fal
    se

    You might also like