Computer Science Speaking Skills Talk

  • Gates Hillman Centers
  • Traffic21 Classroom 6501
  • Ph.D. Student
  • Computer Science Department
  • Carnegie Mellon University
Speaking Skills

ML for ML: Learning Cost Semantics by Experiment

It is an open problem in static resource bound analysis to connect high-level resource bounds (order of complexity) with the actual execution time and memory usage of compiled  machine code. We propose to use machine learning to derive a cost model for a high-level source language that approximates the execution cost of compiled programs, including garbage collection, on a specific hardware platform. The technique has been implemented for a subset of OCaml using Inrias OCaml compiler on an Intel x86-64 and ARM 64-bit v8-A platform with execution time and heap allocations being the considered resources. In this talk, I will define cost semantics, motivate and describe our machine learning approach and finally present the evaluation

Joint work with Jan Hoffman

Presented in Partial Fulfillment of the CSD Speaking Skills Requirement

For More Information, Please Contact: