The Machine Learning Summer Schools (http://mlss.cc) were established in 2002 with the aim of bringing together world class speakers from academia, the national labs, and industry to deliver tutorial-style lectures over a two week period. This project supports the two week long Machine Learning Summer School (MLSS) at UC Santa Cruz, CA during July 9-20 2012.
Intellectual Merits: Machine learning has many important applications in science and industry. Modern machine learning uses a mix of insights from different disciplines, most notably artificial intelligence, statistics and optimization - areas that traditionally have not had much overlap. The participants will take part in tutorials given by experts from several different areas of machine learning - an opportunity that many students do not have at their home institutions. There is substantial integration of research and education in this activity. Furthermore, the participants will be able to interact with the tutorial speakers and with other students.
Broader Impact: The Machine Learning Summer School is designed to enable participants with different backgrounds to gain in-depth knowledge of the current state of the art in machine learning. The participants will interact with leading machine learning experts from academia and industry. It contributes to the creation of a diverse cadre of machine learning researchers and practitioners by offering unique training opportunities for undergraduate and graduate students from under-represented groups through personalized mentoring and scholarships.
In 2002 an International Machine Learning Summer School (MLSS) series was started: www.mlss.cc/. So far 25 summer schools have been conducted. Our goal in 2012 was to increase the participation of US venues in this very effective international effort of educating mostly graduate students in Machine Learning. We conducted a very successful 20th Machine Learning Summer School (MLSS) in Santa Cruz thanks to generous funding from NSF as well as our industrial sponsors: Google, Skytree, IBM, LinkedIn, Pascal2. The summer school was only the 3rd such offering in the USA. However, this year there is one in Carnegie Mellon and next summer there will be one in the University of Texas in Austin. The school in Santa Cruz featured 10 long talks and 12 application talks spread over a two week period (which is the typical length of such summer schools). More details about the school can be found at https://mlss.soe.ucsc.edu/home, and slides and videos of the presentations are linked at https://mlss.soe.ucsc.edu/schedule/speakers. In total we had 112 participants: 2 faculty members, 5 postdoctoral fellows, 8 industrial participants, 93 graduate students and 4 undergraduates. A fraction of 23% of the participants were female, 41% from CA, 38% from states outside of CA and 33% from foreign universities. There were 2 African American students, 6 Arabic students and 1 pacific islander. The students were exposed to a variety of basic and advanced topics in machine learning, many of which are not covered in typical undergraduate and graduate level machine learning classes. Some notable highlights included Prof. Dale Schuurmans who provided an overview of the field and covered supervised, unsupervised and partially-supervised training algorithms, Prof. Vishy Vishwanathan who talked about scaling up Machine Learning algorithms using the latest optimization techniques, and Prof. Yoav Freund who covered a unique view of boosting from an optimization perspective. The students also heard application talks from a wide variety of local companies: Ken Clarkson - IBM, Xavier Amatriain - Netflix, Amnol Bhasin - LinkedIn, Tushar Deepak Chandra, Ravi Kumar, Hartmut Neven and Yoram Singer - Google, Dennis Decoste and Chih-Jen Lin - EBay, Ralf Herbrich - Facebook, Asli Celikyilmaz - Microsoft Research.