Speaker
Description
An amount of the data obtained from the High Luminosity LHC (HL-LHC) accelerator, as well as other future collider experiments, is growing faster. In order to have a good analysis of the data obtained, a high-performance discriminate model is required, and analysis using many machine learning techniques have been studying. However, due to the limitations of the present neural network which are usually trained to possess a poor level of abstraction for rather simple but robust mathematical relations, the quality of conventional classification models are highly dependent on the quality, quantity, and range of the data used for training. In this talk, we introduce a new architecture of classification model, which have not only high performance but also have improved model-universality so that the valid range of the model is highly extended. Finally, we demonstrate a new classification model with interpretable latent space abstracted by our neural machine.
