Global AI and Data Science

Global AI & Data Science

Train, tune and distribute models with generative AI and machine learning capabilities

 View Only

  • 1.  Probability Functor Models

    Posted 22 days ago

    I have published a document on probability functor models on Zenodo. One can define a probability measure and use the resulting probability function as the model function.



    ------------------------------
    John Harby
    CEO
    Autonomic AI, LLC
    Temecula CA
    9513835000
    ------------------------------


  • 2.  RE: Probability Functor Models

    Posted 19 days ago

    Hi John,

    Thank you for sharing this insightful work on probability functor models. As an Associate Professor focusing on psychometrics and statistical modeling, I find your approach to using probability functions as model functions very relevant to our challenges in parameter estimation and latent trait modeling.

    Best regards, Ahmed Megahed



    ------------------------------
    Ahmed
    Megahed
    Zagazig University
    Associate Professor of Educational Psychology
    Asamir@foe.zu.edu.eg
    Zagazig
    Egypt
    ------------------------------

    Original Message


  • 3.  RE: Probability Functor Models

    Posted 19 days ago

    Hi Ahmed,

    Thank you, I really appreciate that perspective. The connection to latent trait modeling is exactly where I think this framework becomes interesting, especially around separating context (projection) from inference and treating probability as domain selection rather than parameter noise.

    I'd be very interested in exploring how this could apply to IRT-style models or identifiability issues.

    Best regards,
    John



    ------------------------------
    John Harby
    CEO
    Autonomic AI, LLC
    Temecula CA
    9513835000
    ------------------------------

    Original Message


  • 4.  RE: Probability Functor Models

    Posted 19 days ago

    Dear John,

    Thank you for your encouraging response. I am thrilled that you see the potential in connecting probability functors with latent trait modeling.

    Regarding your interest in IRT-style models and identifiability, these are precisely the challenges I focus on. In fact, I have just released an updated version (0.2.0) of my R package, 'GLMBasedRaschEstimation', on CRAN. It utilizes a Generalized Linear Model (GLM) framework to address parameter estimation in Rasch models, which aligns with your idea of treating probability as domain selection and separating context from inference.

    I would be very interested in exploring how your 'Probability Functor' framework could provide more robust solutions for model identifiability or perhaps even enhance the estimation algorithms we use in psychometrics.

    Let's keep the dialogue open; I believe there is a significant opportunity for a cross-disciplinary bridge between your AI-driven approach and traditional psychometric modeling.

    Best regards,

    Ahmed Samir Megahed Assistant Professor, Zagazig University



    ------------------------------
    Ahmed
    Megahed
    Zagazig University
    Associate Professor of Educational Psychology
    Asamir@foe.zu.edu.eg
    Zagazig
    Egypt
    ------------------------------

    Original Message


  • 5.  RE: Probability Functor Models

    Posted 15 days ago

    Hi John,

    Thank you for sharing this insightful document on probability functor models. As someone working in educational psychology and psychometrics, I find your approach to using probability functions as model functions very relevant to our work in parameter estimation and latent trait modeling.

    Looking forward to exploring your findings on Zenodo!

    Best regards,



    ------------------------------
    Ahmed Megahed Zagazig University Associate Professor of Educational Psychology Asamir@foe.zu.edu.eg Zagazig Egypt
    ------------------------------

    Original Message


Related Content