Deep Learning (DL) · Advanced level

Advanced level · Module 2 of 6 · 20 guided hours

Generative Models

Understand how autoencoders, GANs, diffusion models and large language models generate content, and assess their risks.

Learning objectives

  • Explain autoencoders and latent representations
  • Describe how GANs, diffusion models and language models generate content
  • Assess risks of generated content and plan safeguards

Topics

  • autoencoders and latent spaces
  • GANs
  • diffusion models
  • large language models and next-token prediction
  • evaluating generated content
  • deepfakes, misinformation and safeguards

Lessons

  1. Lesson 1 Autoencoders and Latent Spaces

    Compress data into a latent representation and reconstruct it, and use reconstruction error to detect anomalies.

    encoder and decoderlatent spacereconstruction lossanomaly detection

  2. Lesson 2 GANs and Diffusion Models

    Understand how generative adversarial networks and diffusion models create realistic images.

    generator and discriminatoradversarial trainingdiffusion and denoisingtext-to-image

  3. Lesson 3 Large Language Models

    Understand how large language models generate text, why they can be wrong and how they are adapted.

    next-token predictionpre-training and instruction tuninghallucinationretrieval augmentation

  4. Lesson 4 Risks and Safeguards for Generative AI

    Assess harms from generated content and design safeguards for generative applications.

    misinformationdeepfakes and consentcontent provenancehuman review

Practical activity

Train a small autoencoder for anomaly detection on sensor images, then evaluate a language model's outputs for a local information task and design safeguards.

Expected outcomes

  • Understanding of how generative models work
  • Responsible assessment of generated content

Assessment

Practical on the autoencoder plus a written risk assessment for a generative application.

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.