Speaker
Weiyi Sun
Description
We present a neural-network-based method for transverse position reconstruction in strip AC-LGADs for future collider applications. Laser scans of a four-strip sensor with different inter-electrode gaps and signal intensities are used to study the reconstruction performance. Neural-network approaches are compared with conventional amplitude-based methods, with particular attention to spatial resolution, generalization to untrained positions, and performance at different signal intensities. Preliminary results show that neural networks can significantly improve the position reconstruction performance, while their extrapolation capability and the information content of different waveform features are being further investigated.
