Particle physics processes are usually modelled with complex Monte-Carlo simulations of the hard process, parton shower, and detector interactions. These simulators typically do not admit a tractable likelihood function: given a (potentially high-dimensional) set of observables, it is usually not possible to calculate the probability of these observables for some model parameters. Particle physicists usually tackle this problem of “likelihood-free inference” by hand-picking a few “good” observables or summary statistics and filling histograms of them. But this conventional approach discards the information in all other observables and often does not scale well to high-dimensional problems.
In the three publications “Constraining Effective Field Theories With Machine Learning”, “A Guide to Constraining Effective Field Theories With Machine Learning”, and “Mining gold from implicit models to improve likelihood-free inference”, a new approach has been developed. In a nut shell, additional information is extracted from the simulators. This “augmented data” can be used to train neural networks to efficiently approximate arbitrary likelihood ratios. We playfully call this process “mining gold” from the simulator, since this information may be hard to get, but turns out to be very valuable for inference.
- Constraining effective field theories with machine learning (Alexander Held, 07 Nov 2019) at 24th International Conference on Computing in High Energy & Nuclear Physics
- Constraining effective field theories with machine learning (Johann Brehmer, 06 Jun 2019) at INFN Padova seminar
- Constraining effective field theories with machine learning (Johann Brehmer, 18 Apr 2019) at Higgs and Effective Field Theory 2019
- 'Mining gold' from simulators to improve likelihood-free inference (Johann Brehmer, 18 Mar 2019) at Likelihood-free inference workshop
- Keynote: Constraining effective field theories with machine learning (Johann Brehmer, 14 Mar 2019) at International Workshop on Advanced Computing and Analysis Techniques in Physics Research
- Bringing together simulations, physics insight, and machine learning to constrain new physics (Johann Brehmer, 28 Feb 2019) at Dark universe seminar
- Meticulous measurements with matrix elements and machine learning (Johann Brehmer, 14 Jan 2019) at ITS/CHEP joint seminar
- Improving inference with matrix elements and machine learning (Johann Brehmer, 11 Jan 2019) at HK IAS Program on High Energy Physics
- Learning to constrain new physics (Johann Brehmer, 20 Sep 2018) at IPPP seminar
- Learning to constrain new physics (Johann Brehmer, 13 Sep 2018) at Pheno & Vino Seminar
- Learning to constrain new physics (Johann Brehmer, 27 Aug 2018) at Elementary particle seminar
- Machine Learning to Probe a BSM Higgs Sector (Johann Brehmer, 23 Jul 2018) at Higgs Hunting
- Constraining Effective Theories with Machine Learning (Johann Brehmer, 27 Jun 2018) at Theory seminar
- The frontier of simulation-based inference, K. Cranmer, J. Brehmer and G. Louppe, arXiv 1911.01429 (Submitted to National Academy of Sciences) (04 Nov 2019).
- Mining for Dark Matter Substructure: Inferring subhalo population properties from strong lenses with machine learning, J. Brehmer, S. Mishra-Sharma, J. Hermans, G. Louppe and K. Cranmer, arXiv 1909.02005 (04 Sep 2019).
- Benchmarking simplified template cross sections in $WH$ production, J. Brehmer, S. Dawson, S. Homiller, F. Kling and T. Plehn, arXiv 1908.06980 (19 Aug 2019).
- MadMiner: Machine learning-based inference for particle physics, J. Brehmer, F. Kling, I. Espejo and K. Cranmer, arXiv 1907.10621 (24 Jul 2019).
- Effective LHC measurements with matrix elements and machine learning, J. Brehmer, K. Cranmer, I. Espejo, F. Kling, G. Louppe et. al., arXiv 1906.01578 (04 Jun 2019).