Name of Reviewer ------------------ Suporn Pongnumkul Key Contribution ------------------ Summarize the paper's main contribution(s). Address yourself to both the class and to the authors, both of whom should be able to agree with your summary. This paper proposed a method for visual category recognition, which improves the recognition rate of the current best result by 3-5%. The approached taken was to learn a globally-consistent image-to-image distance functions using large-margin formulation Novelty -------- Does this paper describe novel work? If you deem the paper to lack novelty please cite explicitly the published prior work which supports your claim. Citations should be sufficient to locate the paper and page unambiguously. Do not cite entire textbooks without a page reference. Is not clear to me what the novelty of this paper is, but I am not an expert in this area. The application of (the relazation of) large-margin classification seems to be most noval. Other steps taken seem to have been done by other work and are steps taken to improve performances. Reference to prior work ----------------------- Please cite explicitly any prior work which the paper should cite. Reference seems good (upto my knowledge) Clarity ------- Does it set out the motivation for the work, relationship to previous work, details of the theory and methods, experimental results and conclusions as well as can be expected in the limited space available? Can the paper be read and understood by a competent graduate student? Are terms defined before they are used? Is appropriate citation made for techniques used? The motivation was set out well and the relationship to previous work was explained. However, more details could be laid out about previous work for easier access and understanding. The numerical algorithms were touched upon briefly. Technical Correctness --------------------- You should be able to follow each derivation in most papers. If there are certain steps which make overly large leaps, be specific here about which ones you had to skip. The flow of the technical part seems good. Experimental Validation ----------------------- For experimental papers, how convinced are you that the main parameters of the algorithms under test have been exercised? Does the test set exercise the failure modes of the algorithm? For theoretical papers, have worked examples been used to sanity-check theorems? Speak about both positive and negative aspects of the paper's evaluation. Since the algorithm was tested on a universal (large) data set, it is convincing that the algorithm is tested well, in both working and failure modes. The recognition rates seem plausible. However, it would be nice to see examples of the failure and hear some possible explanation of the causes. Overall Evaluation ------------------ This paper presents an algorithm for visual category recognition. The paper has enough technical contribution and improves result from the previous methods. However the novelty of this paper was not outlined very well. People who are not experts in the area might have a hard time figuring that out. Questions and Issues for Discussion ----------------------------------- What questions and issues are raised by this paper? What issues do you think this paper does not address well? How can the work in this paper be extended? 1. To understand the "5. Details" section better, it would be nice to see the effects of each subsection (e.g. the performance rates with the present of all triplets compared to the ones with the pruned triplets.) 2. The paper states: "Most of our performance is gained from the shape features; at 15 images per category, the color features add only 2% to the mean recognition. At the same time, the fact that these simple color features improve performance at all demonstrates that our method is able to naturally combine features of very different types." A very interestion future work direction is to see more features combined and see how the performance rate changes.