We present AL, a novel automated machine learning system that learns to generate new supervised learning pipelines from an existing corpus of supervised learning programs. In contrast to existing automated machine learning tools, which typically implement a search over manually selected machine learning functions and classes, AL learns to identify the relevant classes in an API by analyzing dynamic program traces that use the target machine learning library. AL constructs a conditional probability model from these traces to estimate the likelihood of the generated supervised learning pipelines and uses this model to guide the search to generate pipelines for new datasets. Our evaluation shows that AL can produce successful pipelines for datasets that previous systems fail to process and produces pipelines with comparable predictive performance for datasets that previous systems process successfully.
Fri 25 Oct Times are displayed in time zone: Beirut change
14:00 - 14:22 Talk | AL: Autogenerating Supervised Learning Programs OOPSLA DOI | ||
14:22 - 14:45 Talk | Program Synthesis with Algebraic Library Specifications OOPSLA Benjamin MarianoUniversity of Maryland, College Park, Josh ReeseUniversity of Maryland, College Park, Siyuan XuPurdue University, ThanhVu NguyenUniversity of Nebraska, Lincoln, Xiaokang QiuPurdue University, Jeffrey S. FosterTufts University, Armando Solar-LezamaMassachusetts Institute of Technology DOI | ||
14:45 - 15:07 Talk | AutoPandas: Neural-Backed Generators for Program Synthesis OOPSLA Rohan BavishiUC Berkeley, Caroline LemieuxUniversity of California, Berkeley, Roy FoxUC Berkeley, Koushik SenUniversity of California, Berkeley, Ion StoicaUC Berkeley DOI | ||
15:07 - 15:30 Talk | On the Fly Synthesis of Edit Suggestions OOPSLA Anders MiltnerPrinceton University, Sumit GulwaniMicrosoft, Vu LeMicrosoft, Alan LeungMicrosoft, Arjun RadhakrishnaMicrosoft, Gustavo SoaresMicrosoft, Ashish TiwariMicrosoft, Abhishek UdupaMicrosoft DOI Pre-print Media Attached |