Doing research in machine learning can be challenging! Consequently, we will be able to
work with only a small number of students, and enrollment will be limited.
One of the goals of the class is to help students get involved in
long-term research. Specifically, it is targeted to students who anticipate being at
Stanford until at least December 2010. Thus, for example, freshmen, sophomore
or junior students are eligible to take the class, as are seniors who have
already been admitted to a co-term program. First year MSc students and PhD
students are also eligible, but not senior students, co-term students or second
year MSc students who are graduating in June 2010.
To quickly get you the background knowledge you'll need to do research in deep
learning, all students are required to successfully complete a programming
assignment on deep learning (posted below) by Wednesday
January 13th. This programming assignment asks you to implement the sparse
autoencoder algorithm, described in the lecture
notes. If you are a good C++ programmer, and have taken and mastered the
material in CS221 or CS229, we believe you should be able to successfully
complete this assignment. This programming assignment will also give you a
lower-bound on the pace you can expect in CS294A.
If you are already involved in AI research, or if you already have experience with
deep learning algorithms (for example, if you had previously already
implemented a sparse autoencoder), you may be exempt from these requirements.
Please contact Prof. Ng if you would like to be considered for an exemption.