Enhanced AdaBoost with Adaptive Weighting for Higgs Signal Classification

The 2012 Higgs boson discovery at CERN highlighted the difficulty of isolating signals within petabytes of Large Hadron Collider data. This study introduces an innovative AdaBoost extension, building on Freund and Schapire’s framework, with a novel adaptive weighting scheme a' _m that tackles class imbalance and high-dimensionality in classifying Higgs boson collisions from the UCI HIGGS dataset. Achieving 70.40% test accuracy, the model leverages an exponential loss function to prioritize challenging data points, optimizing signal detection while balancing overfitting. Results peak at depth 3 and 98 iterations, showcasing enhanced performance over standard methods. This breakthrough demonstrates AdaBoost’s potential for large-scale physics classification, with future refinements like dynamic classifier families proposed to elevate accuracy further.