WG211/M19Shan - Revision history

Chung-chieh: /* From high-level inference algorithms to efficient code */ Add joint authors

2019-04-15T16:42:28Z

From high-level inference algorithms to efficient code: Add joint authors

← Older revision		Revision as of 18:42, 15 April 2019
Line 1:		Line 1:
	== From high-level inference algorithms to efficient code ==		== From high-level inference algorithms to efficient code ==

			(joint work with Rajan Walia, Jacques Carette, Praveen Narayanan, and Sam Tobin-Hochstadt)

	Probabilistic programming languages are valuable because they allow		Probabilistic programming languages are valuable because they allow

Chung-chieh at 16:41, 15 April 2019

2019-04-15T16:41:34Z

← Older revision		Revision as of 18:41, 15 April 2019
Line 1:		Line 1:
	# From high-level inference algorithms to efficient code		== From high-level inference algorithms to efficient code ==

	Probabilistic programming languages are valuable because they allow		Probabilistic programming languages are valuable because they allow

Chung-chieh at 16:41, 15 April 2019

2019-04-15T16:41:03Z

← Older revision		Revision as of 18:41, 15 April 2019
Line 1:		Line 1:
	* From high-level inference algorithms to efficient code		# From high-level inference algorithms to efficient code

	Probabilistic programming languages are valuable because they allow		Probabilistic programming languages are valuable because they allow

Chung-chieh: Add Shan abstract

2019-04-15T16:40:52Z

Add Shan abstract

New page

* From high-level inference algorithms to efficient code

Probabilistic programming languages are valuable because they allow
domain experts to express probabilistic models and inference algorithms
without worrying about irrelevant details. However, for decades there
remained an important and popular class of probabilistic inference
algorithms whose efficient implementation required manual low-level
coding that is tedious and error-prone. They are algorithms whose idiomatic expression requires
random array variables that are ''latent'' or whose likelihood is
''conjugate''. Although that is how practitioners
communicate and compose these algorithms on paper, executing such
expressions requires ''eliminating'' the latent variables and
''recognizing'' the conjugacy by symbolic mathematics. Moreover,
matching the performance of handwritten code requires speeding up loops
by more than a constant factor.

We show how probabilistic programs that directly and concisely express
these desired inference algorithms can be compiled
while maintaining efficiency. We introduce new transformations that
turn high-level probabilistic programs with arrays into pure loop code.
We then make great use of domain-specific invariants and norms to
optimize the code, and to specialize and JIT-compile the code per execution.
The resulting performance is competitive with manual implementations.