Combined W H → lvbb search at Cdf with Neural Network and Matrix Element techniques

We present a Standard Model (S M) Higgs search at Cdf for W H → l n b b produced by p p collisions at a center-of-mass energy of √s = 1.96 TeV using an integrated luminosity of 4.3 fb-1. This new analysis uses several improvements to increase sensitivity for Higgs searches including a Neural Network (N N) b-tagger, a b-jet energy correction which improves Higgs dijet mass resolution, isolated track selection to increase lepton acceptance, a N N flavor separator to reduce mistags, and a three-jet category to complement the standard two-jet procedure. We perform two independent analyses with distinct multivariate discriminant techniques, N N analysis and Matrix Element Technique (M E). The N N analysis sets limits across a range of Higgs masses from 100 GeV/c2 to 150 GeV/c2, with observed limits ranging from σ(p p → W±H)x B R(H → b b) < 4.0 to 37.6 times the S M prediction and 5.3 x S M observed(4.0 expected) at M(H) = 115 GeV/c2. The M E analysis sets limits with observed limits ranging from σ (p p→W±H)x B R(H → b b) < 4.5 to 48.8 times the S M prediction and 6.6 x S M observed (4.1 expected) at M(H) = 115 GeV/c2. For the Cdf combination we use two-jet events from N N analysis and three-jet events from M E analysis together with other Cdf searches. We observe no excess in data and we calculate combined upper limits on the ratio of the Higgs boson cross section times the branching ratio to its S M prediction for Higgs boson masses between 100 and 200 GeV/c2. The results are in good agreement with the expectations obtained from background-only pseudo-experiments. The observed (expected) 95% C L upper limit is 3.1 (2.4) higher than the S M production cross section for Higgs boson mass of 115 GeV/c2.