Course Projects

Each student attending the 2000 summer course will be expected to complete a computational project during the two week session, and these projects can be the collaborative work of up to two students. We feel that projects are an important component of the course for two reasons. First, they serve as focal points for discussion and interaction between the students and faculty. Second, they allow each student to examine a single issue in depth and to use the tools presented in the course to develop that idea. In fact, many of the projects developed by students in previous years have become significant research projects and later published papers. To encourage depth and quality in the projects we ask all students to make a very informal presentation of their project at the end of the course. Presentations typically take the form of a poster-like session on the last day of the course and are very informal to discourage students from taking the projects TOO seriously.

Below are just some ideas for possible projects. As new ideas for projects arise during the course we will add them to this list. In addition to looking at the tutorials listed after each project you may want to look at the box of papers we have brought along for related material or you may even want to talk to one of us to learn about related work. Finally, the main campus has an excellent library that you can use, although you may want to talk to one of the organizers to find out how to use the photocopier there.

  1. Light Adaptation. As you know, light adaptation has been studied psychophysically and physiologically at several levels. Any of those are interesting systems to model computationally. You might begin by having a look at the linSysTutorial.
  2. Ganglion Cell Physiology. After hearing Fred Rieke and EJ talk you might want to consider modeling the behavior of single RGCs or small networks of RGCs. To prepare for this project you might look at the linSysTutorial and the imageTutorial.
  3. Selectivity and gain control in V1: orientiation selectivity, direction selectivity, simple and complex cells, cross-orientation inhibition, surround supression/facilitation. The relevant tutorials will include: linSysTutorial, imageTutorial, pyramidTutorial, and V1MTTutorial.
  4. MT physiology. Basically everything and anything about MT physiology has been modeled. Many of the most famous culprits are right here. Wanna' try your hand at this? It actually is a pretty interesting thing to do (even Paul thinks so). The relevant tutorials include: linSys, image, pyramid, motion, and V1MT.
  5. Non-Fourier Motion. These are sometimes called higher-order motion. They are stimuli perceived by humans as moving but not analyzed as "motion" by linear motion analyzers. You might even be able to think of a stimulus that we could call fourth-order motion. Relevant tutorials include: linSys, image, pyramid, motion, and V1MT.
  6. Pattern Vision. Modeling of processes associated with pattern detection and discrimination, contrast appearance/matching, lateral masking and facilitation. Relevant tutorials include: linSys, pyramid, V1MT, sdt, and masking.
  7. Image representation and image statistics. This is probably one of the hotest topics in vision right now and forms the core of one of Eero's lectures. Lots of related material can be found on Eero's homepage (www.cns.nyu.edu/~eero). (No, Eero didn't write this.) The relevant tutorials include: linSys, image, pyramid, motion, and V1MT.
  8. Texture. Barlow meets Julesz, a steel cage grudge match. Lots of people have wondered how, or if, texture patterns are encoded in the visual system. Again, Eero's website may be useful for this one. Relevant tutorials include: linSys, sampling, image, pyramid, V1MT, and masking.
  9. Color. The first step is to figure out what color space is. Some people do, others go their whole lives avoiding the color sessions at ARVO. This is your big chance to actually understand color constancy or color appearance and to be, as Brainard says, a color weenie. Brainard and EJ will get you started with their lectures. Tutorials: svd, imageFormation, colorMatching, and color.
  10. Synapses and spiking. This is an opportunity to think seriously about the roles of noise, spike generation reliability, neural codes, and synaptic reliability. Shadlen is a super expert on this and you might have a look at his website. Relevant tutorials: poisson, sdt, and intAndFire.
  11. Non-Traditional Color Opponency. Can you design an alternive to Red-Green and Blue-Yellow opponency? Cyan-Purple? Can you find unique aqua? The other 23 students might make entirely unbiased observers to use in your experiments. EJ and Brainard ought to be able to help get you started and if you succeed, well, ah, EJ'll see that you never work in this town again. Relevant tutorials: colorMatching, and color.