[Illinois] MCB 493 Lecture 5: Unsupervised Learning and Distributed Representations

By Thomas J. Anastasio

Department of Molecular and Integrative Physiology , University of Illinois at Urbana-Champaign, Urbana, IL

View Presentation

Additional materials available (3)

Category

Online Presentations

Published on

Abstract

Unsupervised learning algorithms, given only a set of input patterns, can train neural networks to form distributed representations of those patterns that resemble brain maps.

5.1 Learning through Competition to Specialize for Specific Inputs

5.2 Training Few Output Neurons to Represent Many Input Patterns

5.3 Simulating the Formation of Brain Maps using Cooperative Mechanisms

5.4 Modeling the Formation of Tonotopic Maps in the Auditory System

5.5 Simulating the Development of Orientation Selectivity in the Visual Cortex

5.6 Modeling a Possible Multisensory Map in the Superior Colliculus

Bio

THOMAS J. ANASTASIO SYSTEMS NEUROBIOLOGIST I am an associate professor in the University of Illinois Department of Molecular and Integrative Physiology and a full-time faculty member of the Beckman Institute. My main research interest is in Systems Neurobiology, which I define as the application of conceptual and computational methods to complex and multilevel problems in neurobiology. I am also involved in teaching, writing, and technology transfer.

Cite this work

Researchers should cite this work as follows:

  • Thomas J. Anastasio (2013), "[Illinois] MCB 493 Lecture 5: Unsupervised Learning and Distributed Representations," https://nanohub.org/resources/16951.

    BibTex | EndNote

Time

Location

University of Illinois at Urbana-Champaign, Urbana, IL

Submitter

NanoBio Node, Obaid Sarvana, George Daley

University of Illinois at Urbana-Champaign

Tags

  1. Illinois
  2. learning
  3. NanoBio Node at Illinois
  4. Neural Systems Modeling
  5. neuroscience
  6. tonotopic maps
  7. unsupervised learning
  8. visual systems
[Illinois] MCB 493 Lecture 5: Chapter 5
  • Figure 5.1 Reponses of a 1. Figure 5.1 Reponses of a "face… 0
    00:00/00:00
  • Figure 5.1 Reponses of a 2. Figure 5.1 Reponses of a "face… 416.77615927419356
    00:00/00:00
  • Table 5.1 3. Table 5.1 527.8503528225807
    00:00/00:00
  • Figure 5.2 A generic, two-layered neural network 4. Figure 5.2 A generic, two-laye… 580.66018145161286
    00:00/00:00
  • Table 5.2 5. Table 5.2 1223.5516633064517
    00:00/00:00
  • Table 5.3 6. Table 5.3 1432.0637096774194
    00:00/00:00
  • Table 5.4 7. Table 5.4 1741.7326108870966
    00:00/00:00
  • Table 5.5 8. Table 5.5 1901.1538306451612
    00:00/00:00
  • Figure 5.3 Competitive learning results in specialization 9. Figure 5.3 Competitive learnin… 2029.335433467742
    00:00/00:00
  • Table 5.6 10. Table 5.6 2131.7319556451612
    00:00/00:00
  • Figure 5.4 A simple example of vector quantization 11. Figure 5.4 A simple example of… 2133.0955897177419
    00:00/00:00
  • Figure 5.5 A slightly updated view of the homunculus in the human somatosensory cortex 12. Figure 5.5 A slightly updated … 2263.260660282258
    00:00/00:00
  • Table 5.7 13. Table 5.7 2563.012222782258
    00:00/00:00
  • Figure 5.6 Specialization without map formation using purely competitive learning 14. Figure 5.6 Specialization with… 2729.2516129032256
    00:00/00:00
  • Figure 5.7 Specialization and map formation using combined competitive and cooperative learning 15. Figure 5.7 Specialization and … 2784.1688760080647
    00:00/00:00
  • Figure 5.8 Map of isofrequency bands within the dorsal cochlear nucleus of the cat 16. Figure 5.8 Map of isofrequency… 3000.7387600806451
    00:00/00:00
  • Figure 5.9 Types of input unit tuning curves 17. Figure 5.9 Types of input unit… 3099.9121471774197
    00:00/00:00
  • Figure 5.10 The responses of a tonotopically ordered set of broadly tuned input units 18. Figure 5.10 The responses of a… 3189.2921622983868
    00:00/00:00
  • Figure 5.11 Responses of the output units in a 20-by-10 network trained using the SOM to form a tonotopic map 19. Figure 5.11 Responses of the o… 3435.4900957661289
    00:00/00:00
  • Figure 5.12 The two-dimensional input pattern arrays used in the original model of self-organization in the primary visual (striate) cortex 20. Figure 5.12 The two-dimensiona… 3760.1589717741936
    00:00/00:00
  • Figure 5.13 The preferred orientations of output units in a model of the striate cortex following training using the self-organization algorithm of von der Malsburg 21. Figure 5.13 The preferred orie… 3822.8861391129035
    00:00/00:00
  • Table 5.8 22. Table 5.8 4062.01796875
    00:00/00:00
  • Table 5.9 23. Table 5.9 4062.01796875
    00:00/00:00
  • Table 5.10 24. Table 5.10 4062.01796875
    00:00/00:00
  • Figure 5.14 Orientation selectivity and columnar organization in the primary visual cortex 25. Figure 5.14 Orientation select… 4062.01796875
    00:00/00:00
  • Figure 5.15 A modality map as it might occur in the superior colliculus 26. Figure 5.15 A modality map as … 4062.01796875
    00:00/00:00
Pop Out