Deep Learning (DL) · Intermediate level

Intermediate level · Module 5 of 6 · 13 guided hours

Evaluating Deep Learning Models

Evaluate deep models rigorously, analyse their errors and diagnose underfitting and overfitting.

Learning objectives

  • Design validation strategies that reflect real use
  • Analyse errors systematically
  • Diagnose and fix underfitting and overfitting

Topics

  • validation strategies and test sets
  • metrics per class
  • error analysis by category
  • confidence and calibration
  • underfitting vs overfitting
  • robustness checks

Lessons

  1. Lesson 1 Validation Strategies

    Choose validation and test sets that reflect how the model will really be used.

    in-distribution vs field testper-class recallmacro-averaged metricstest-set discipline

  2. Lesson 2 Error Analysis

    Look systematically at what the model gets wrong and why, to guide improvements.

    error categoriesconfusion pairsconfidently wrong predictionsfix priorities

  3. Lesson 3 Diagnosing Underfitting and Overfitting

    Use training and validation curves to decide whether to make the model bigger, regularise it or get more data.

    training vs validation curvesunderfitting remediesoverfitting remediesdata vs model problems

Practical activity

Evaluate the medicine package models on a separately collected 'field' test set, analyse errors by condition and propose improvements.

Expected outcomes

  • Honest evaluation that reflects field use
  • Targeted improvement plans

Assessment

Evaluation report marked on methodology, error analysis and diagnosis.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.