1 / 17

ECE 599/692 – Deep Learning Lecture 10 – Regularized AE and Case Studies

Explore the use of regularized autoencoders in deep learning and examine case studies in data representation and feature engineering. Learn how to solve the overfitting problem and improve performance in machine learning algorithms.

helizabeth
Download Presentation

ECE 599/692 – Deep Learning Lecture 10 – Regularized AE and Case Studies

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. ECE 599/692 – Deep LearningLecture 10 – RegularizedAEandCaseStudies Hairong Qi, Gonzalez Family Professor Electrical Engineering and Computer Science University of Tennessee, Knoxville http://www.eecs.utk.edu/faculty/qi Email: hqi@utk.edu

  2. Outline • Lecture 9: Points crossed • General structure of AE • Unsupervised • Generative model? • The representative power • Basic structure of a linear autoencoder • Denoising autoencoder (DAE) • AE in solving overfitting problem • Lecture 10: RegularizedAEandcasestudies • Lecture11:VAEleadingtoGAN

  3. Theperformanceofmachinelearningmethodsisheavilydependentonthechoiceofdatarepresentation(orfeatures)onwhichtheyareapplied.…muchoftheactualeffortindeployingmachinelearningalgorithmsgoesintothedesignofpreprocessingpipelinesanddatatransformationsthatresultinarepresentationofthedatathatcansupporteffectivemachinelearning.…Suchfeatureengineeringisimportantbutlabor-intensiveandhighlightstheweaknessofcurrentlearningalgorithms:theirinabilitytoextractandorganizethediscriminativeinformationfromthedata….Theperformanceofmachinelearningmethodsisheavilydependentonthechoiceofdatarepresentation(orfeatures)onwhichtheyareapplied.…muchoftheactualeffortindeployingmachinelearningalgorithmsgoesintothedesignofpreprocessingpipelinesanddatatransformationsthatresultinarepresentationofthedatathatcansupporteffectivemachinelearning.…Suchfeatureengineeringisimportantbutlabor-intensiveandhighlightstheweaknessofcurrentlearningalgorithms:theirinabilitytoextractandorganizethediscriminativeinformationfromthedata…. [Bengio:2014]

  4. Differentapproaches • Probabilisticmodels • Capturestheposteriordistributionoftheunderlyingexplanatoryfactorsgivenobservations • Reconstruction-basedalgorithms • AE • Geometricallymotivatedmanifold-learningapproaches

  5. Issues Requirestrongassumptionsofthestructureinthedata Makesevereapproximations,leadingtosuboptimalmodels RelyoncomputationallyexpensiveinferenceprocedureslikeMCMC

  6. BasicstructureofAE z W2 y W1 x

  7. RegularizedAE–SparseAE z y …… x

  8. RegularizedAE–DenoisingAE?[DAE:2008] Thewell-knownlinkbetween“trainingwithnoise”andregularization

  9. Application Example: AE in Spectral Unmixing

  10. Spectral Unmixing [Bannon, Nature Photonics 2009] [1] L. Miao, H. Qi, “Endmember extraction from highly mixed data using minimum volume constrained non-negative matrix factorization,” IEEE Transactions on Geoscience and Remote Sensing, 45(3):765-777, March 2007. (Highest Impact Paper Award) [2] L. Miao, H. Qi, H. Szu, “A maximum entropyapproach to unsupervised mixed pixel decomposition,” IEEE Transactions on Image Processing, 16(4):1008-1021, April 2007. • Mathematic formulation of HSI mixing: • x: sensor readout • A: source matrix • s: abundance vector • n: measuring noise • Constraints • Sum-to-one • Nonnegative • Unsupervised unmixing • MVC-NMF • GDME

  11. Unmixing and DL? Deep Learning [Ciznicki,SPIE 2012] Spectral Unmixing A good marriage? [CVPR 2012 Tutorial by Honglak Lee]

  12. AutoEncoder (AE) How to formulate the unmixing problem in the framework of AutoEncoder? What is endmember? What is abundance? How to incorporate the constraints? What are the constraints? • X = AS vs. ^X = As(W1X) • Part-based AE assumes • Tied weight: W1 = AT • Nonnegative weight: [1]Chen, Minmin, et al. "Marginalized denoising autoencoders for domain adaptation." arXiv preprint arXiv:1206.4683 (2012). [2] Lemme, Andre, René Felix Reinhart, and Jochen Jakob Steil. "Efficient online learning of a non-negative sparse autoencoder." ESANN. 2010.

  13. AutoEncoder Cascade Rui Guo, Wei Wang, Hairong Qi, “Hyperspectral image unmixing using cascaded autoencoder,” IEEE Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensor (WHISPERS), Tokyo, Japan, June 2-5, 2015. (Best Paper Award) • Marginalized Denoising AutoEncoder (mDA) • Non-negative Sparse AutoEncoder (NNSA)

  14. Discussion on the Network Structure X = AS = As(W1X) = W1Ts(W1X) That is, S = s(W1X) On tied weight

  15. Discussion on the Network Structure (cont’d) X = AS = As(W1X) Assume s is linear: AW1=I On the nonnegative constraint

  16. Proposed Algorithm T

  17. Proposed Algorithm • Objective function: • The denoising constraint: • multilayer scaled marginalized denosing autoencoder (mDA): Wn • Nonnegativity: • Sum-to-one:

More Related