Tag Archives: fast algorithm

Divide-and-Conquer Learning by Anchoring a Conical Hull

Many well-known machine learning methods aim to draw a line between two classes. However, in our recently accepted NIPS 2014 paper “Divide-and-Conquer Learning by Anchoring a Conical Hull“, we reduce lots of fundamental machine learning problems (a broad class of … Continue reading

Posted in Tianyi's work | Tagged , , , , , , | Leave a comment

Multi-task Copula – A semiparametric joint prediction model for multiple outputs with sparse graph structure

Our paper “Multi-task Copula by Sparse Graph Regression“ has been accepted by KDD 2014 this year. So we can talk at the conference which is at NYC, between August 24-27. Before that, let me introduce this new method. In summary, we tackle … Continue reading

Posted in Tianyi's work | Tagged , , | Leave a comment

NeSVM (Nesterov’s method for SVM) code for our ICDM 2010 paper

You can now download MATLAB code for NeSVM from here.  In the code, options.mu is a key parameter to adjust the trade-off between consistent decreasing of primal object function, and the speed. So you need to roughly tune it to … Continue reading

Posted in Tianyi's work | Tagged , , | Leave a comment

AISTATS 2013 GreBsmo code is released

Here is the GreBsmo code for our AISTATS 2013 paper. You can use it as a greedy version of GoDec solver for X=L+S problem. It is much faster and more robust. There are three video subsequences you can play in … Continue reading

Posted in Tianyi's work | Tagged , , , , | Leave a comment

[Best student paper award] Welcome to my “Divide-and-Conquer Anchoring (DCA)” talk at ICDM Dallas Dec 8

Is it possible to finish a 60000×10000 matrix decomposition (NMF, PCA, etc)  or completion in 6 seconds on your laptop’s matlab? Can we make it even faster by a simple distributable scheme? How to summarize a huge-scale dataset (ratings, movie, … Continue reading

Posted in Tianyi's work | Tagged , , , , , , | Leave a comment

Our DMKD paper is selected as Top 5 Editor’s Choice Article for Free Reading

Prof. Geoff Webb, the Editor-in-Chief of Data Mining and Knowledge Discovery (Springer) announced in his kdnuggets website that our paper “Manifold Elastic Net: A Unified Framework for Sparse Dimension Reduction”, which was published on DMKD journal in 2011 and cited … Continue reading

Posted in Tianyi's work | Tagged , , , , , | 1 Comment

Greedy Bilateral (GreB) Paradigm for Large-scale Matrix Completion, Robust PCA and Low-rank Approximation

Our paper “Greedy Bilateral Sketch, Completion and Smoothing” has been accepted by AISIATS 2013. Abstracts reads below, PDF is here, and code will be coming soon. Abstract: Recovering a large low-rank matrix from highly corrupted, incomplete or sparse outlier overwhelmed … Continue reading

Posted in Tianyi's work | Tagged , , , , , | 3 Comments

Compressed Labeling: An important extension of Hamming Compressed Sensing; at NIPS now

We are just informed that our submission “Compressed Labeling (CL) on Distilled Labelsets (DL) for Multi-label Learning” is accepted by Machine Learning Journal (Springer). Online first PDF can be downloaded here. CL is an important application and extension of Hamming … Continue reading

Posted in Tianyi's work | Tagged , , , , , , , | Leave a comment

Semi-Soft GoDec: >4 times faster, auto-determined k

Here is a good news of GoDec (pertaining to our ICML 2011 paper): Semi-Soft GoDec is released. Different from the ordinary GoDec which imposes hard threshholding to both the singular values of the low-rank part L and the entries of the … Continue reading

Posted in Tianyi's work | Tagged , , , , , , , | 2 Comments

Hamming Compressed Sensing-recovering k-bit quantization from 1-bit measurements with linear non-iterative algorithm

We developed a new compressed sensing type signal acquisition paradigm called “Hamming Compressed Sensing (HCS)” to recover signal’s k-bit quantization rather than itself. Directly recovering quantization is much more preferred in practical digital systems. HCS provides a linear, non-iterative quantization … Continue reading

Posted in Tianyi's work | Tagged , , , | 3 Comments