How to model Factors and its influence that converge or diverge resulting in the bias of ML/AI outcomes.

- Selection Bias, Algorithmic Bias, Training Data Bias, Information bias, Anchoring bias, Confirmation, Stability bias, Cognitive bias of Coders etc.

  1. Characterisation of ML/AI driven data collection, models and codebase to effectively detect bias
  2. ML anti-bias object modelling in GSIM
  3. Code and Data Bias in Interpretable Vs Blackbox ML/AI models 
  4. AI/ML Bias - The role of making decisions from experience (DFE) (Cognitive and/Or Pre-Trained experience Models)
  5. Application of modelling Thin-slicing concept in the detection of implicit and explicit bias in ML/AI  (Process objects, Data and Codebase)

1 Comment

  1. Krishnan Ambady

    18 Jun, 2021
    • a Good* background reading : https://towardsdatascience.com/the-actual-difference-between-statistics-and-machine-learning-64b49f07ea3
    • Iterated Algorithmic Bias in the Interactive Machine Learning Process of Information Filtering
      ack : Wenlong Sun1, , Olfa Nasraoui1, and Patrick Shafto2
      • Dept of Computer Engineering and Computer Science, University of Louisville, Louisville, KY, U.S.A.
        Dept of Mathematics and Computer Science, Rutgers University - Newark, Newark, NJ, U.S.A
    • Prediction of Job Performance : Thin slices have been found to be predictive of performance in a number of analysis domains : For example, judgments of teachers’ personality characteristics from thin-slice clips accurately predicted the teachers’ end-of-semester ratings by students and principals (Ambady & Rosenthal, 1993). Three 10-sec silent video clips of each of 13 university teachers were rated by nine naïve raters on 15 variables: accepting, active, anxious, attentive, competent, confident, dominant, empathic, enthusiastic, honest, likeable, optimistic, professional, supportive, and warm.Teacher effectiveness was appraised through course evaluations at the end of the semester and was correlated with a composite variable composed of 14 of the ratings (all but anxious). .. read further details ... on https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.476.4572&rep=rep1&type=pdf