Olshausen’s Demo. 1.The Training set ?  Natural Images (Olhausen’s database)  How much do we learn ?  face database and car database 2.The Sparseness.

Slides:



Advertisements
Similar presentations
Spike Based Visual Encoding Activity level (a m ) Visual encoder implemented in the NEF as network of 1024 laterally inhibiting neural columns Network.
Advertisements

Weakly supervised learning of MRF models for image region labeling Jakob Verbeek LEAR team, INRIA Rhône-Alpes.
Face Recognition: A Convolutional Neural Network Approach
Advanced topics.
Generalizing Backpropagation to Include Sparse Coding David M. Bradley and Drew Bagnell Robotics Institute Carnegie.
Learning Representations. Maximum likelihood s r s?s? World Activity Probabilistic model of neuronal firing as a function of s Generative Model.
Patch-based Image Deconvolution via Joint Modeling of Sparse Priors Chao Jia and Brian L. Evans The University of Texas at Austin 12 Sep
Facial feature localization Presented by: Harvest Jang Spring 2002.
Multiple People Detection and Tracking with Occlusion Presenter: Feifei Huo Supervisor: Dr. Emile A. Hendriks Dr. A. H. J. Stijn Oomes Information and.
DISCRIMINATIVE DECORELATION FOR CLUSTERING AND CLASSIFICATION ECCV 12 Bharath Hariharan, Jitandra Malik, and Deva Ramanan.
Freeman Algorithms and Applications in Computer Vision Lihi Zelnik-Manor Lecture 5: Pyramids.
Genetic Algorithms  An example real-world application.
MSU CSE 803 Stockman Fall 2009 Vectors [and more on masks] Vector space theory applies directly to several image processing/representation problems.
Discussion: Urban terrain segmentation for the Marmara Region Speaker: Akarun Discussant: Lerner-Lam.
Challenges in Computer Vision Understanding the ”seeing machine” The input (images) The output (shapes, actions?, diagnosis?) The mapping (statistics,
Active Appearance Models Computer examples A. Torralba T. F. Cootes, C.J. Taylor, G. J. Edwards M. B. Stegmann.
M.Sc. CNS Visual Perception Concept of receptive field Stan Gielen Dept. of Biophysics.
Extraction of Salient Contours in Color Images Vonikakis Vasilios, Ioannis Andreadis and Antonios Gasteratos Democritus University of Thrace
ICA Alphan Altinok. Outline  PCA  ICA  Foundation  Ambiguities  Algorithms  Examples  Papers.
Artificial Neural Networks
SOS Boosting of Image Denoising Algorithms
Autoencoders Mostafa Heidarpour
1 MSU CSE 803 Fall 2014 Vectors [and more on masks] Vector space theory applies directly to several image processing/representation problems.
Overview of Back Propagation Algorithm
Face Recognition Using Neural Networks Presented By: Hadis Mohseni Leila Taghavi Atefeh Mirsafian.
CIVS, Statistics Dept. UCLA Deformable Template as Active Basis Zhangzhang Si UCLA Department of Statistics Ying Nian Wu, Zhangzhang Si, Chuck.
Multiclass object recognition
Sparse Coding Arthur Pece Outline Generative-model-based vision Linear, non-Gaussian, over-complete generative models The penalty method.
Online Learning for Matrix Factorization and Sparse Coding
Multimodal Interaction Dr. Mike Spann
Creating With Code.
Unsupervised Learning of Compositional Sparse Code for Natural Image Representation Ying Nian Wu UCLA Department of Statistics October 5, 2012, MURI Meeting.
Presented by: Mingyuan Zhou Duke University, ECE June 17, 2011
Projects: 1.Predictive coding in balanced spiking networks (Erwan Ledoux). 2.Using Canonical Correlation Analysis (CCA) to analyse neural data (David Schulz).
Representations for object class recognition David Lowe Department of Computer Science University of British Columbia Vancouver, Canada Sept. 21, 2006.
Machine Learning, Decision Trees, Overfitting Machine Learning Tom M. Mitchell Machine Learning Department Carnegie Mellon University January 14,
Fields of Experts: A Framework for Learning Image Priors (Mon) Young Ki Baik, Computer Vision Lab.
Ψ Sensation & Perception.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Adaptive FIR Neural Model for Centroid Learning in Self-Organizing.
Understanding early visual coding from information theory By Li Zhaoping Lecture at EU advanced course in computational neuroscience, Arcachon, France,
Object and face recognition
Hidden Variables, the EM Algorithm, and Mixtures of Gaussians Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem 02/22/11.
Robodog Frontal Facial Recognition AUTHORS GROUP 5: Jing Hu EE ’05 Jessica Pannequin EE ‘05 Chanatip Kitwiwattanachai EE’ 05 DEMO TIMES: Thursday, April.
NMF Demo: Lee, Seung Bryan Russell Computer Demonstration.
Computer Vision: 3D Shape Reconstruction Use images to build 3D model of object or site 3D site model built from laser range scans collected by CMU autonomous.
Independent Component Analysis features of Color & Stereo images Authors: Patrik O. Hoyer Aapo Hyvarinen CIS 526: Neural Computation Presented by: Ajay.
© Stocktrek Images/Getty Images 1 Web Search What is bioluminescence? 2 Web Search Name a few creatures that have bioluminescent properties. 3.
9.012 Presentation by Alex Rakhlin March 16, 2001
Spontaneous activity in V1: a probabilistic framework
Article Review Todd Hricik.
Energy Preserving Non-linear Filters
Learning Objective Using the generative form of Bayes’ equation, the learning objective is to find the most probable explanation, H, for the input patterns,
Tableau Overview  Tableau is widely used data visualization and BI tool. Tableau is simple to use and has extensive visualization capability that make.
Mean transform , a tutorial
Face Recognition with Deep Learning Method
Given the series: {image} and {image}
Bryan Russell Computer Demonstration
Feature Selection Analysis
Binary Image processing بهمن 92
network of simple neuron-like computing elements
Sparselet Models for Efficient Multiclass Object Detection
Question: how are neurons in the primary visual cortex encoding the visual scene?
Vision: In the Brain of the Beholder
Adaboost for faces. Material
Vinit Shah, Joseph Picone and Iyad Obeid
Both series are divergent. A is divergent, B is convergent.
Example segmentations - unseen images
Face Recognition: A Convolutional Neural Network Approach
Determine whether the sequence converges or diverges. {image}
Outline Announcement Neural networks Perceptrons - continued
Presentation transcript:

Olshausen’s Demo

1.The Training set ?  Natural Images (Olhausen’s database)  How much do we learn ?  face database and car database 2.The Sparseness term ?  Prior steepness  Sparseness function 3.Natural encoding or hacking?  Whitening the data  Non-stationary hypothesis How Important Is:

Training with Natural Images  Training: 10 images (512x512)  10,000 presentations  Batch size: 100  Basis Function: 16x16

Face Database  Training: 100 images (100x100)  10,000 presentations  Batch size: 100  Basis Function: 16x16

Encoding Properties Original 50 basis 10 basis 30 basis40 basis 20 basis

Car Database  Training: 200 images (128x128)  10,000 presentations  Batch size: 100  Basis Function: 16x16

Comments 1.The algorithm seems to capture the structure of the images (cf car):  Learning is experience-dependent 2.Basis functions found in good agreement with properties of neurons in visual cortex:  Receptive fields are localized, oriented, bandpass

1.The Training set ?  Background, face and car databases 2.The Sparseness term ?  Prior steepness  Sparseness function 3.Natural encoding or hacking?  Whitening the data  Non-stationary hypothesis How Important Is:

Prior Steepness Steepness 2.2 Steepness 10 Steepness 5 Steepness 100

Prior Steepness Steepness 2.2Steepness 1.5 Steepness 0.2

Sparseness Function

S(x)=|x| S(x)=log(1+x^2)

Sparseness Function batch of 100 samples: Mean Error: abs=.471 / log =.504

1.The Training set ?  Background, face and car databases 2.The Sparseness term ?  Prior steepness  Sparseness function 3.Natural encoding or hacking?  Whitening the data  Non-stationary hypothesis How Important Is:

Whitening the Data Data are filtered with whitening/low-pass filter: How important is it for the convergence of the algorithm? The question is to know whether it is just a speed-up or is it required for convergence?

Non-preprocessed Car Images  Training: 100 images (100x100)  30,000 presentations  Batch size: 100  Basis Function: 16x16

Non-stationary Hypothesis: Encoding the Full Face After few iterations…

Code + images available: