It's also true that I'm not exactly a genius when it comes to these subjects, so my ideas might be completely misled. This approach doesn't seem the most ideal to me given the situation I'm dealing with, using log-loss / cross-entropy as the loss function seems a more sane approach.
by using squared error as its loss function, as noted in the docs of the python library I'm working with. However, Funk SVD (and subsequent improvements) all seem to be built to work on a traditional item rating scale (1-5), and not on binary data i.e. One of the most famous methods to decompose the matrix is using Funk SVD. My ultimate goal is to achieve a meaningful representation of the users in order to cluster them.īy scouting around the web I have learned that a good approach to do so is applying matrix factorization on my original user-item matrix, and then run the clustering algorithm on the resulting user's matrix. In said matrix, 1 represents a positive feedback, -1 a negative one, and 0 means the user has never seen the product. Thus, the matrix looks like so: user_id | item_1 | item_2 | item_3 |. However, instead of the usual 1 to 5 rating scale, items can only receive a positive (1) or negative (-1) feedback. Shi, "An Objective Distortion Measure for Binary Document Images Based on Human Visual Perception", in Proceedings of the Sixteenth International Conference on Pattern Recognition (ICPR 2002), Quebec City, vol. Shi, "Distance-Reciprocal Distortion Measure for Binary Document Images", IEEE Signal Processing Letters, Vol. Kot and Jun Cheng, "Secure Data Hiding in Binary Document Images for Authentication", in Proceedings of the 2003 IEEE International Symposium on Circuits and Systems (ISCAS 2003), Bangkok, Thailand, vol. Kot and Lihui Chen, "Watermark Embedding in DC Components of DCT for Binary Images", in Proceedings of the IEEE International Workshop on Multimedia Signal Processing (MMSP 2002), US Virgin Islands, pp. Kot and Lihui Chen, "Binary Image Watermarking through Blurring and Biased Binarization", International Journal of Image and Graphics Special Issue on Image Data Hiding, Vol. FAQS.ORG reserves the right to edit your answer as to improve its clarity. If youd like to get expert points and benefit from positive ratings, please create a new account or login into an existing account below. Your answer will not be displayed immediately. Watermark Embedding in DC Components of DCT for Binary ImagesĤa. Your answer will be published for anyone to see and rate. Kot and Rahardja Susanto, "Binary Image Watermarking through Biased Binarization", in Proceedings of the 2003 IEEE International Conference on Multimedia & Expo (ICME 2003), Baltimore, Maryland, vol.
Binary Image Watermarking through Biased Binarizationģa. The Distance-Reciprocal Distortion MeasureĢa.
Binary images for testing, including full page CCITT test images, theirĬropped versions, and some other binary images.ġa. ***In all documents and papers reporting research work that uses the matlabĬodes provided here, please cite the respective paper(s).
However, by referring to the respective papers, I hope you can get most of them.
Due to time constraint, the codes are not well-documented. Test images, and codes for binary image watermarking (or data hiding) algorithms Singapore, on binary image watermarking (or data hiding). This Matlab package shares my earlier work at Nanyang Technological University,