On an optimal use of tags: CS, MF, ML, implementation, AMP and BlindDeconvolution and more....

It has come to my attention that I may not be doing an optimal job at providing the right tag for the right blog entry. In particular, I am sometimes not putting the most appropriate label and while putting another less used one. 

For this reason, I'll try to be consistent with regards to certain labels and subjects areas (you are welcome to let me know about tag failures and if specific tags ought to be added, this specific blog entry has been tagged with all these tags in order to let interested people know about it): 

• When the topic is about Compressive Sensing, I am going to use CS from now on. 
• When the topic talks about Matrix Factorization, I'll use MF only from now on
• When the topic goes into Machine Learning territory, I'll use ML
• When the blog entry features an implementation, I use the implementation tag (there are 333 so far !)
• When the topic discusses how to infer a forward model for a sensing system using incomplete information on either said forward model or the input that is fed into that model, or both then I'll used BlindDeconvolution. That topic is very diverse as it includes some aspects of calibration, deblurring, blind compressive sensing, network discovery....This is a tough call here because really when the unknown forward model is nonlinear and coarse, we could call this whole endeavor instances of Machine Learning (Supervised and Unsupervised Learning) while in other cases, it absolutely fit into the generic Matrix Factorization category. For the time being, since those communities have not really talking to each other (it is just a question of time), we will use BlindDeconvolution tag. There are already 60 entries on the subject, here some of the ones from the past two years:

• Self-Calibration and Biconvex Compressive Sensing
• Efficient Blind Compressed Sensing and $\ell_0$ Sparsifying Transform Learning
• Random Calibration for Accelerating MR-ARFI Guided Ultrasonic Focusing in Transcranial Therapy 
• Lifting for Blind Deconvolution in Random Mask Imaging: Identifiability and Convex Relaxation / Learning to Deblur
• Non-parametric PSF estimation from celestial transit solar images using blind deconvolution  
• Compressive Hyperspectral Imaging with Side Information
• MONGOOSE: An exact arithmetic toolbox for a consistent and reproducible structural analysis of metabolic network models
• Sparse Representation Issues in Hyperspectral and Multispectral Images
• Euclid in a Taxicab: Sparse Blind Deconvolution with Smoothed $\ell_1/ell_2$ Regularization
• Sunday Morning Insight: Zero Knowledge Sensor Design / Data Driven Sensor Design
• Cal-AMP : Blind Sensor Calibration using Approximate Message Passing 
• Enhancing Pure-Pixel Identification Performance via Preconditioning - implementation - 
• Reconstructing Nonlinear Biochemical Networks
• Blind Image Deblurring by Spectral Properties of Convolution Operators - implementation
• Balancing Sparsity and Rank Constraints in Quadratic Basis Pursuit - implementation -
• CaSPIAN: A Causal Compressive Sensing Algorithm for Discovering Directed Interactions in Gene Networks - implementation - 
• Hyperspectral Imaging: Hardware and Data Reconstruction
• Blind Image Deblurring with Unknown Boundaries Using the Alternating Method of Multipliers - implementation -
• Blind Identification via Lifting - implementation -
• Robust Sparse Signal Recovery for Compressed Sensing with Sampling and Representation Uncertainties - implementation - 
• Convex Optimization Approaches for Blind Sensor Calibration using Sparsity /A Conjugate Gradient Algorithm for Blind Senor Calibration in Sparse Recovery
• SAHD: Dynamic L1 Reconstruction - Justin Romberg 
• The Fukushima Inverse Problem - implementation -
• Blind Calibration in Compressed Sensing using Message Passing Algorithms
• Analysis Based Blind Compressive Sensing - implementation -
• Blind Sensor Calibration in Sparse Recovery Using Convex Optimization and Analysis Based Blind Compressive Sensing
• Compressive System Identification (CSI): Theory and Applications of Exploiting Sparsity in the Analysis of High-Dimensional Dynamical Systems
• Blind compressive sensing dynamic MRIRobust Locally Linear Analysis with Applications to Image Denoising and Blind Inpainting -implementation - 
• Coded Hyperspectral Imaging and Blind Compressive Sensing
• Correcting Camera Shake by Incremental Sparse Approximation - implementation -
• Phase Diagram and Approximate Message Passing for Blind Calibration and Dictionary Learning
• Sunday Morning Insight: The extreme paucity of tools for blind deconvolution of biochemical networks
• Blind Deconvolution using Convex Programming - implementation -

Same thing goes for AMP solvers with blog entries using the AMP tag as well as the following tags:

• CSHardware (92)
• CSjobs (85)
• QuantCS (62)
• phaseretrieval (59)
• SaturdayMorningVideos (58)
• hyperspectral (58)
• nonlinearCS (56)
• SundayMorningInsight (55)
• CSVideo (47)
• RandNLA (47)
• grouptesting (44)
• tensor (44)
• 1bit (41)
• NuitBlancheReview (30)
• Csstats (29)
• RandomFeatures (29)
• MLParis (22)
• ImagingWithNature (17)
• SAHD (17)
• mapmaker (15)
• nanopore (15)
• phasediagrams (8)
• ICLR2015 (11)
• GreatThoughtsFriday (7)
• CitingNuitBlanche (6)
• HammingsTime (6)
• accidentalcamera (5)
• CompressiveSensingWhatIsItGoodFor (4)
• BaltiAndBioinformatics (1)

 

Image Credit: NASA/JPL/Space Science Institute : Full-Res: W00091207.jpg
W00091207.jpg was taken on January 08, 2015 and received on Earth January 09, 2015. The camera was pointing toward SATURN, and the image was taken using the IR2 and IR1 filters.

  Join the CompressiveSensing subreddit or the Google+ Community and post there !
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.