Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model. Baydin, A. G., Heinrich, L., Bhimji, W., Gram-Hansen, B., Louppe, G., Shao, L., Cranmer, K., Wood, F., & others In *Thirty-second Conference on Neural Information Processing Systems (NeurIPS)*, 2019.

Link Paper abstract bibtex

Link Paper abstract bibtex

We present a novel probabilistic programming framework that couples directly to existing large-scale simulators through a cross-platform probabilistic execution protocol, which allows general-purpose inference engines to record and control random number draws within simulators in a language-agnostic way. The execution of existing simulators as probabilistic programs enables highly interpretable posterior inference in the structured model defined by the simulator code base. We demonstrate the technique in particle physics, on a scientifically accurate simulation of the tau lepton decay, which is a key ingredient in establishing the properties of the Higgs boson. Inference efficiency is achieved via inference compilation where a deep recurrent neural network is trained to parameterize proposal distributions and control the stochastic simulator in a sequential importance sampling scheme, at a fraction of the computational cost of a Markov chain Monte Carlo baseline.

@inproceedings{BAY-19a, title={Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model}, author={Baydin, Atilim Gunes and Heinrich, Lukas and Bhimji, Wahid and Gram-Hansen, Bradley and Louppe, Gilles and Shao, Lei and Cranmer, Kyle and Wood, Frank and others}, booktitle={Thirty-second Conference on Neural Information Processing Systems (NeurIPS)}, archiveprefix = {arXiv}, eprint = {1807.07706}, year={2019}, url_Link={https://papers.nips.cc/paper/8785-efficient-probabilistic-inference-in-the-quest-for-physics-beyond-the-standard-model}, url_Link={https://arxiv.org/abs/1807.07706}, url_Paper={https://papers.nips.cc/paper/8785-efficient-probabilistic-inference-in-the-quest-for-physics-beyond-the-standard-model.pdf}, url_Paper={https://arxiv.org/pdf/1807.07706.pdf}, abstract={We present a novel probabilistic programming framework that couples directly to existing large-scale simulators through a cross-platform probabilistic execution protocol, which allows general-purpose inference engines to record and control random number draws within simulators in a language-agnostic way. The execution of existing simulators as probabilistic programs enables highly interpretable posterior inference in the structured model defined by the simulator code base. We demonstrate the technique in particle physics, on a scientifically accurate simulation of the tau lepton decay, which is a key ingredient in establishing the properties of the Higgs boson. Inference efficiency is achieved via inference compilation where a deep recurrent neural network is trained to parameterize proposal distributions and control the stochastic simulator in a sequential importance sampling scheme, at a fraction of the computational cost of a Markov chain Monte Carlo baseline.} }

Downloads: 0

{"_id":"3gxqHD2ivMGD37MeA","bibbaseid":"baydin-heinrich-bhimji-gramhansen-louppe-shao-cranmer-wood-etal-efficientprobabilisticinferenceinthequestforphysicsbeyondthestandardmodel-2019","authorIDs":[],"author_short":["Baydin, A. G.","Heinrich, L.","Bhimji, W.","Gram-Hansen, B.","Louppe, G.","Shao, L.","Cranmer, K.","Wood, F.","others"],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","title":"Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model","author":[{"propositions":[],"lastnames":["Baydin"],"firstnames":["Atilim","Gunes"],"suffixes":[]},{"propositions":[],"lastnames":["Heinrich"],"firstnames":["Lukas"],"suffixes":[]},{"propositions":[],"lastnames":["Bhimji"],"firstnames":["Wahid"],"suffixes":[]},{"propositions":[],"lastnames":["Gram-Hansen"],"firstnames":["Bradley"],"suffixes":[]},{"propositions":[],"lastnames":["Louppe"],"firstnames":["Gilles"],"suffixes":[]},{"propositions":[],"lastnames":["Shao"],"firstnames":["Lei"],"suffixes":[]},{"propositions":[],"lastnames":["Cranmer"],"firstnames":["Kyle"],"suffixes":[]},{"propositions":[],"lastnames":["Wood"],"firstnames":["Frank"],"suffixes":[]},{"firstnames":[],"propositions":[],"lastnames":["others"],"suffixes":[]}],"booktitle":"Thirty-second Conference on Neural Information Processing Systems (NeurIPS)","archiveprefix":"arXiv","eprint":"1807.07706","year":"2019","url_link":"https://arxiv.org/abs/1807.07706","url_paper":"https://arxiv.org/pdf/1807.07706.pdf","abstract":"We present a novel probabilistic programming framework that couples directly to existing large-scale simulators through a cross-platform probabilistic execution protocol, which allows general-purpose inference engines to record and control random number draws within simulators in a language-agnostic way. The execution of existing simulators as probabilistic programs enables highly interpretable posterior inference in the structured model defined by the simulator code base. We demonstrate the technique in particle physics, on a scientifically accurate simulation of the tau lepton decay, which is a key ingredient in establishing the properties of the Higgs boson. Inference efficiency is achieved via inference compilation where a deep recurrent neural network is trained to parameterize proposal distributions and control the stochastic simulator in a sequential importance sampling scheme, at a fraction of the computational cost of a Markov chain Monte Carlo baseline.","bibtex":"@inproceedings{BAY-19a,\n title={Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model},\n author={Baydin, Atilim Gunes and Heinrich, Lukas and Bhimji, Wahid and Gram-Hansen, Bradley and Louppe, Gilles and Shao, Lei and Cranmer, Kyle and Wood, Frank and others},\n booktitle={Thirty-second Conference on Neural Information Processing Systems (NeurIPS)},\n archiveprefix = {arXiv},\n eprint = {1807.07706},\n year={2019},\n url_Link={https://papers.nips.cc/paper/8785-efficient-probabilistic-inference-in-the-quest-for-physics-beyond-the-standard-model},\n url_Link={https://arxiv.org/abs/1807.07706},\n url_Paper={https://papers.nips.cc/paper/8785-efficient-probabilistic-inference-in-the-quest-for-physics-beyond-the-standard-model.pdf},\n url_Paper={https://arxiv.org/pdf/1807.07706.pdf},\n abstract={We present a novel probabilistic programming framework that couples directly to existing large-scale simulators through a cross-platform probabilistic execution protocol, which allows general-purpose inference engines to record and control random number draws within simulators in a language-agnostic way. The execution of existing simulators as probabilistic programs enables highly interpretable posterior inference in the structured model defined by the simulator code base. We demonstrate the technique in particle physics, on a scientifically accurate simulation of the tau lepton decay, which is a key ingredient in establishing the properties of the Higgs boson. Inference efficiency is achieved via inference compilation where a deep recurrent neural network is trained to parameterize proposal distributions and control the stochastic simulator in a sequential importance sampling scheme, at a fraction of the computational cost of a Markov chain Monte Carlo baseline.}\n}\n\n\n","author_short":["Baydin, A. G.","Heinrich, L.","Bhimji, W.","Gram-Hansen, B.","Louppe, G.","Shao, L.","Cranmer, K.","Wood, F.","others"],"key":"BAY-19a","id":"BAY-19a","bibbaseid":"baydin-heinrich-bhimji-gramhansen-louppe-shao-cranmer-wood-etal-efficientprobabilisticinferenceinthequestforphysicsbeyondthestandardmodel-2019","role":"author","urls":{" link":"https://arxiv.org/abs/1807.07706"," paper":"https://arxiv.org/pdf/1807.07706.pdf"},"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"inproceedings","biburl":"https://raw.githubusercontent.com/plai-group/bibliography/master/group_publications.bib","creationDate":"2020-07-03T21:04:56.076Z","downloads":0,"keywords":[],"search_terms":["efficient","probabilistic","inference","quest","physics","beyond","standard","model","baydin","heinrich","bhimji","gram-hansen","louppe","shao","cranmer","wood","others"],"title":"Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model","year":2019,"dataSources":["7avRLRrz2ifJGMKcD","BKH7YtW7K7WNMA3cj","wyN5DxtoT6AQuiXnm"]}