Close Menu
    Trending
    • Credit Risk Scoring for BNPL Customers at Bati Bank | by Sumeya sirmula | Jul, 2025
    • The New Career Crisis: AI Is Breaking the Entry-Level Path for Gen Z
    • Musk’s X appoints ‘king of virality’ in bid to boost growth
    • Why Entrepreneurs Should Stop Obsessing Over Growth
    • Implementing IBCS rules in Power BI
    • What comes next for AI copyright lawsuits?
    • Why PDF Extraction Still Feels LikeHack
    • GenAI Will Fuel People’s Jobs, Not Replace Them. Here’s Why
    AIBS News
    • Home
    • Artificial Intelligence
    • Machine Learning
    • AI Technology
    • Data Science
    • More
      • Technology
      • Business
    AIBS News
    Home»Machine Learning»Supervised Learning: Kernel Methods — CS229 | by Shreyansh Jain | Jan, 2025
    Machine Learning

    Supervised Learning: Kernel Methods — CS229 | by Shreyansh Jain | Jan, 2025

    Team_AIBS NewsBy Team_AIBS NewsJanuary 24, 2025No Comments4 Mins Read
    Share Facebook Twitter Pinterest LinkedIn Tumblr Reddit Telegram Email
    Share
    Facebook Twitter LinkedIn Pinterest Email


    On this article i’ll be sharing an understanding and mathematical facet of Kernel Strategies just about Stanford College’s CS229 course taught by the famend British-American pc scientist Dr.Andrew Ng and Dr. Tengyu Ma.

    Function maps

    Typically, a easy linear perform could not give correct outcomes for an issue. For instance, if we’re predicting home costs, a straight-line relationship (linear perform) won’t seize the sample properly. non-linear perform would possibly match the information higher and provides extra correct predictions.

    We are able to see a linear perform fails right here

    Think about a cubic equation:

    We are able to rewrite this perform utilizing a function map ϕ : R→R⁴ which maps a single-dimensional actual quantity R right into a four-dimensional vector R⁴ :

    Let θ ∈ R⁴ be the vector containing θ0,θ1,θ2,θ3 as entries. Then we are able to

    rewrite the cubic perform in x as:

    x is the unique enter attribute, and ϕ(x) are the function variables

    In such circumstances, we want fashions that may deal with extra complicated relationships. Kernel strategies assist by remodeling the information right into a higher-dimensional area utilizing a kernel perform, making it simpler for the mannequin to seek out patterns and make higher predictions.

    LMS (least imply squares) with options

    In fundamental linear regression, the mannequin is represented by θ^T x , the place x represents the enter options and θ is the vector of parameters to be discovered. Nonetheless, when a easy linear mannequin just isn’t ample to seize the connection, we remodel the enter information utilizing a function map ϕ(x) , which maps the enter x right into a higher-dimensional area. The purpose is then to suit the mannequin θ^T *ϕ(x) , the place θ is a vector of parameters within the higher-dimensional area, making it extra able to capturing complicated relationships. For odd least squares, the batch gradient descent replace is:

    Up to date batch gradient descent for odd least squares

    Equally stochastic gradient descent replace rule is:

    LMS with the Kernel Trick

    When coping with high-dimensional function mappings, direct computation turns into costly, particularly when the function map entails high-degree polynomials. To beat this, we are able to use a kernel trick that avoids explicitly computing the function map. As a substitute, we use a kernel perform Ok(x, z) , which computes the inside product

    within the higher-dimensional area. This enables for extra environment friendly computation, because the kernel perform will be precomputed and doesn’t require the specific computation of ϕ(x) . The replace rule for the parameters θ will be expressed as a linear mixture of the function vectors, and we symbolize θ iteratively with coefficients β. The replace rule for β depends upon the kernel values, and the kernel perform captures the required info from the function map. This ends in an environment friendly gradient descent algorithm that avoids explicitly working with high-dimensional function vectors, enhancing each time and area complexity. The ultimate prediction will be computed utilizing the kernel values, while not having the specific type of ϕ(x) .

    The gradient descent replace rule will be expressed as a linear mixture of the function vectors, represented as

    At every iteration, the replace for the coefficients βi is given by:

    The kernel trick permits us to carry out operations within the higher-dimensional area with out explicitly computing ϕ(x) , making the method extra computationally environment friendly. Lastly, the prediction will be computed as:

    Legitimate Kernels and the Mercer Theorem

    Mercer’s Theorem states that for a perform Ok(x, z) to be a legitimate kernel, it should correspond to an inside product in a higher-dimensional function area. Legitimate kernels should fulfill two situations: symmetry (i.e., Ok(x, z) = Ok(z, x) ) and constructive semi-definiteness (guaranteeing the Gram matrix is non-negative). Some common kernels embrace:



    Source link

    Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
    Previous ArticleQuantum TikTok + Memecoin Mania + Chris Hayes on the Attention Wars
    Next Article Multi-Headed Cross Attention — By Hand | by Daniel Warfield | Jan, 2025
    Team_AIBS News
    • Website

    Related Posts

    Machine Learning

    Credit Risk Scoring for BNPL Customers at Bati Bank | by Sumeya sirmula | Jul, 2025

    July 1, 2025
    Machine Learning

    Why PDF Extraction Still Feels LikeHack

    July 1, 2025
    Machine Learning

    🚗 Predicting Car Purchase Amounts with Neural Networks in Keras (with Code & Dataset) | by Smruti Ranjan Nayak | Jul, 2025

    July 1, 2025
    Add A Comment
    Leave A Reply Cancel Reply

    Top Posts

    Credit Risk Scoring for BNPL Customers at Bati Bank | by Sumeya sirmula | Jul, 2025

    July 1, 2025

    I Tried Buying a Car Through Amazon: Here Are the Pros, Cons

    December 10, 2024

    Amazon and eBay to pay ‘fair share’ for e-waste recycling

    December 10, 2024

    Artificial Intelligence Concerns & Predictions For 2025

    December 10, 2024

    Barbara Corcoran: Entrepreneurs Must ‘Embrace Change’

    December 10, 2024
    Categories
    • AI Technology
    • Artificial Intelligence
    • Business
    • Data Science
    • Machine Learning
    • Technology
    Most Popular

    What’s Behind Meta’s Makeover Ahead of Trump’s Second Term?

    January 8, 2025

    What They Can Teach Us About Communication

    June 14, 2025

    Codes That Can Be Written More Correctly | by AI With Lil Bro | Apr, 2025

    April 26, 2025
    Our Picks

    Credit Risk Scoring for BNPL Customers at Bati Bank | by Sumeya sirmula | Jul, 2025

    July 1, 2025

    The New Career Crisis: AI Is Breaking the Entry-Level Path for Gen Z

    July 1, 2025

    Musk’s X appoints ‘king of virality’ in bid to boost growth

    July 1, 2025
    Categories
    • AI Technology
    • Artificial Intelligence
    • Business
    • Data Science
    • Machine Learning
    • Technology
    • Privacy Policy
    • Disclaimer
    • Terms and Conditions
    • About us
    • Contact us
    Copyright © 2024 Aibsnews.comAll Rights Reserved.

    Type above and press Enter to search. Press Esc to cancel.