Scribd Logo
Upload

Welcome to Scribd!

  • 20 views

    Feature Selection

    Uploaded by

    Javada Javada
    This document discusses feature selection and feature extraction techniques for dimensionality reduction in machine learning. It covers filter-based, wrapper-based and embedded methods for feature selection, and describes the general steps of feature selection procedures.

    Copyright:

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

    Feature Selection

    Uploaded by

    Javada Javada
    20 views18 pages
    This document discusses feature selection and feature extraction techniques for dimensionality reduction in machine learning. It covers filter-based, wrapper-based and embedded methods for feature selection, and describes the general steps of feature selection procedures.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses feature selection and feature extraction techniques for dimensionality reduction in machine learning. It covers filter-based, wrapper-based and embedded methods for feature selection, and describes the general steps of feature selection procedures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    20 views18 pages

    Feature Selection

    Uploaded by

    Javada Javada
    This document discusses feature selection and feature extraction techniques for dimensionality reduction in machine learning. It covers filter-based, wrapper-based and embedded methods for feature selection, and describes the general steps of feature selection procedures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 18

    MODULE 3

    Feature Selection and Feature Extraction

    Feature Selection

    A type of dimensionality reduction where a large number of


    pixels of the image are efficiently represented in such a way that
    interesting parts of the image are captured effectively.

    Dimensionality reduction refers to techniques for reducing the


    number of input variables in training data. When dealing with
    high dimensional data, it is often useful to reduce the
    dimensionality by projecting the data to a lower dimensional
    subspace which captures the “essence” of the data.
    Feature selection is also known as variable selection.
    It uses
    ● data preprocessing techniques of data mining
    ● And is mainly used for declining data by eradicating some
    attribute values.
    ● This process progresses the performance of prediction,
    ● decreases the training time of the algorithm, and
    ● Also offers better imagining of data.
    Feature selections have many application areas such as
    healthcare, ensemble technique, embedded method, etc.

    Feature selection is an approach of selecting the most relevant


    subset features for developing a robust AI/machine learning
    model.

    In the feature selection process,


    ● Redundant and/or irrelevant data is removed from the main
    database.
    ● Performance of the diagnostics model may be improved.
    ● Computational burden on the machine will be reduced.
    ● Computational efficiency is increased.
    Generally, in feature selection procedure four main steps should
    be involved as given:
    (1) subset generation;
    (2) evaluation of the subset;
    (3) procedure stopping criteria; and
    (4) validation.
    In step#1, sub-sets are selected based on the search approach.
    generally, the approach depends on search direction and search
    methodology.
    Step#2 depends on several evaluation parameters such as
    distance, dependency, consistency, etc.
    In the step#3, stopping criteria depends on several other criteria
    (i.e., error is less than required/chosen, complete the search,
    etc.).
    In step#4, the validation of selected attributes is performed using
    different advanced AI/ML algorithms.

    There are different types of feature selection methods. There


    are three main categories of feature selection approaches.
    These approaches are:
    (1) Filter-based Feature Selection (FbFS)
    FbFS is the preprocessing step of any AI/machine learning
    approach. The FbFS method measures the quality (correlation,
    similarity, information, and dependency) of the dataset. Due to
    this, FbFS methods are fast and free from high computational
    calculations.
    Filter-based feature selection chooses the most relevant features
    regardless of the learning algorithm. These methods rely on
    measuring properties of the features, such as correlation,
    entropy, or intra/inter-class distance to determine how
    informative they are.
    (2) Wrapper-based Feature Selection (WbFS)
    WbFS is based on predictive performance of predefined
    predictors. WbFS is not economical in comparison to FbFS
    methods. Some of the approaches under WbFS methods are:
    ● Sequential Selection Algorithms (SSA)
    ● Heuristic-Based Selection Algorithms (HbSA).

    Wrapper-based feature selection uses the learning algorithm to


    evaluate the goodness of a subset of features. This approach has
    the advantage of allowing the detection of likely interaction
    among characteristics, albeit with a higher computation time
    than filter-based methods.
    (3) Embedded Model based Feature Selection (EMbFS) or
    Intrinsic Feature Selection (IFS)
    Embedded-based feature selection aims to combine the
    advantage of both filter and wrapper methods. The feature
    selection is embedded in the learning process of the algorithm.
    For instance, decision trees, such as CART, have a built-in
    mechanism to perform feature selection.
    In The EMbFS category of feature selection, the following
    approaches come in the list as: LASSO, classification &
    regression tree (CART), C4.5, and SVM-RFE, MSVMs, etc.
    Original Features 25 is dropped to 11 Features now below

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy