COMS E6998

1- Fun hands-on Lab 2- Research project (8-page final paper to be submitted by end of semester) + a presentation - we worked in teams. 3- Two presentations: One about a language of your choice, and one about paper set of your choice 4- Weekly Paper Write Ups: 1-page summary every week, 4-papers per week.

3 Assignments, each covering main topic in the class, as mentioned above. In actuality, it feels more like 6 assignments. The B-Spline one will be split into 3 parts, while the geometry processing is split into 2 parts. However, they aren't equal in length. Some will be shorter; others will be longer. 1 Midterm. 1 Final. Totally reasonable and manageable.

1 final project, 2 reports about how your work for the final project is going.

A useful course in terms of the material being applied to software development and industry. However, the assessments in the course are very poor, requiring little effort, but with arbitrary outcomes. First, the quizzes are very detail-oriented and require quickly searching through the books - with lots of subtle tricks. Second, the project is graded quite harshly, for example in our class the average was 80 with 75 low and 85 high, lowering everyone's grade as it was a big part of the grade. So don't expect a great grade.

3 programming assignments(30%). The first one was building a document retrieval system, which was the most time consuming of the three. The second one is sentiment analysis for movie reviews, which was a competition on Kaggle. Your rank on Kaggle would count towards the grade, but don't worry if your rank is not very satisfactory, as the impact it has on the grade is very small. The third one was actually not a programming assignment, and it was the easiest and lightest of the three. 1 final project(30%), can be in the form of a research paper or a software system. Final(30%), like I said above, please do spend time reviewing the materials thoroughly, especially pay attention to ALL those formulas and matrix computations, and make sure you understand how they work.

Fair...The three "programming assignments" are acceptable. Attendances are taken every week. However, the final project is the most confusing part.

Remarkably, at times disturbingly, little. Two problem sets, both assigned near the end of the course and requiring more time to typeset in LaTeX than they did to solve. The main event is the final project, but even that is pretty low-key: the writeup is only six pages long if you're working by yourself, and Vinod seemed surprised when I gave him something in the form of a journal article rather than a lab report. I don't know if MIT students are all crybabies, or if Collins feels compelled to dumb himself down for us (I hope it's the former); either way, don't let the 6 in front of the course number fool you into thinking that the material will make you sweat.

Workload was unexpectedly heavy -- about 1 homework per week (excluding breaks and holidays). It's always hands-on: you get a bunch of binaries to pull apart and answer questions. The time to complete each one was highly variable: some weeks I was done in 3 hours, other times it took me over 12 hours to figure it out. They expect a bit of detail when you answer the questions. The midterm was not very useful. Too many questions exclusively tested whether you remembered random bits of information given in lecture, like what the PE magic number stands for (it's the creator's initials), or whether ASCII and Unicode strings are printed separately by a particular program. There was one short reverse engineering problem on the exam and it was trivial. There's both a final project and a final exam. The project is a more open-ended version of the homework that you can work on with a partner. You have to write a report and you don't get leading questions to point you in the right direction.

This was definitely one of the most fun times I had at Columbia. I took this class without prior knowledge in Machine Translation, but I had some exposure to classifiers and machine learning. Still, Professor Nizar made this class really fun and laid-back. The hands-on lab was a great way to introduce the students to the SMT world and get them excited about the course. The class format hasn't changed much from the last review (I took the class last year). The workload was also almost the same, additionally we were asked to write up a summary of the papers presented every week. But don't let this intimidate you, in fact, this exercise forced us to actually read the papers before the class and it made listening to the presentations more fun and the discussions richer. Also even though some papers were advanced-level and difficult to read but that will not really affect your grade, you will still learn something out of them if you attempt to read them. Overall, this was really fun and laid-back class. Definitely consider it!

A really interesting computer graphics class with an amazing professor. In this class, we are given a glimpse to the realm of geometry modeling and geometry processing. As we live in the age that artists and product designers utilize 3D software such as Rhino or Maya or ZBrush, to digitally model their products before actualization into tangible objects, or to model cartoon characters that would appear in animated films, this class provides a great introduction into the technical aspects of how these 3D softwares work or should work. We study the fundamentals down to the level of so-called NURBS curves and surfaces, geometry processing performed on triangle meshes, and subdivision surfaces -- each with extensive mathematical backgrounds that consider how these structures give artists and manufacturers computational advantages such as desired degree of smoothness no matter what shape of the curves/surfaces are formed. It is a really fun experience to learn how design softwares are actually made of, and we also had a guest speaker from the industry to discuss how he actually uses the software to design real-world products. So the class covers mainly three parts: the fundamentals of B-Spline curves and surfaces (as well as NURBS and Rational form), geometry processing (in particular, mesh data structures and mesh simplification algorithms), and Subdivision surfaces (e.g., Catmull…