After detection, parameter estimation for compact binary signals poses
the most significant computational challenge to their analysis.
Traditional stochastic sampling methods have been deployed for recent
analyses from the O1-O3 runs, but the increasing number of sources
and complexity of waveforms make further efficiency gains essential.
Novel techniques making use of deep learning have been shown to
dramatically reduce latency, at the cost of up-front training, but
will they prove flexible enough for non-standard analyses?
In this talk I will explore the future requirements for parameter estimation,
and outline some of the novel methods that are being used to address
Back to Workshop IV: Big Data in Multi-Messenger Astrophysics