Download presentation

Presentation is loading. Please wait.

Published byAddison Warren Modified over 2 years ago

1
Multi-Label Prediction via Compressed Sensing By Daniel Hsu, Sham M. Kakade, John Langford, Tong Zhang (NIPS 2009) Presented by: Lingbo Li ECE, Duke University 01-22-2010 * Some notes are directly copied from the original paper.

2
Outline Introduction Preliminaries Learning Reduction Compression and Reconstruction Empirical Results Conclusion

3
Introduction Large database of images; Goal: predict who or what is in a given image Samples: images with corresponding labels is the total number of entities in the whole database. One-against-all algorithm: Learn a binary predictor for each label (class). Computation is expensive when is large. e.g., Assume the output vector is sparse.

4
Introduction 111111111111 Main idea: “Learn to predict compressed label vectors, and then use sparse reconstruction algorithm to recover uncompressed labels from these predictions” Compressed sensing : For any sparse vector, it is highly possible to compress to logarithmic in dimension with perfect reconstruction of.

5
Preliminaries : input space; : output (label) space, where Training data: Goal: to learn the predictor with low mean- squared error Assume is very large; Expected value is sparse, with only a few non-zero entries.

6
Learning reduction Linear compression function where Goal: to learn a predictor Predict the label y with the Predictor F Predict the compressed label Ay with the Predictor H Samples Compressed Samples To minimize

7
Reduction-training and prediction Reconstruction Algorithm R: If is close to, then should be close to

8
Compression Functions Examples of valid compression functions:

9
Reconstruction Algorithms Examples of valid reconstruction algorithms: iterative and greedy algorithms Orthogonal Matching Pursuit (OMP) Forward-Backward Greedy (FoBa) Compressive Sampling Matching Pursuit (CoSaMP)

10
General Robustness Guarantees What if the reduction create a problem harder to solve than the original problem? Sparsity error is defined as where is the best k-sparse approximation of

11
Linear Prediction If there is a perfect linear predictor of, then there will be a perfect linear predictor of :

12
Experimental Results Experiment 1: Image data (collected by the ESP Game) 65k images, 22k unique labels; Keep the 1k most frequent labels; the least frequent occurs 39 times while the most frequent occurs about 12k times, 4 labels on average per image; Half of the data as training and half as testing. Experiment 2: Text data (collected from http://delicious.com/) 16k labeled web page, 983 unique labels; the least frequent occurs 21 times, the most frequent occurs about 6500 times, 19 labels on average per web page; Half of the data as training and half as testing. Compression function A: select m random rows of the Hadamard matrix. Test the greedy and iterative reconstruction algorithm: OMP, FoBa, CoSaMp and Lasso. Use correlation decoding (CD) as a baseline method for comparisons.

13
Experimental Results Measure Measure the precision Top two: image data; Bottom: text data

14
Conclusion Application of compressed sensing to multi-label prediction problem with output sparsity; Efficient reduction algorithm with the number of predictions equal to logarithmic in original labels; Robustness Guarantees from compressed case to the original case; and vice versa for the linear prediction setting.

Similar presentations

Presentation is loading. Please wait....

OK

PARALLEL FREQUENCY RADAR VIA COMPRESSIVE SENSING

PARALLEL FREQUENCY RADAR VIA COMPRESSIVE SENSING

© 2018 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on area of parallelogram with vertices Ppt on scheduling in production management Ppt on recycling of solid waste Ppt on nervous system of human body Ppt on marketing management for mba Ppt on solar air conditioning Ppt on time management in marathi Ppt on domain and range Ppt on national education day nepal Ppt on accounting standard 17