Abstract With the growing popularity of artificial intelligence (AI) used for scientific applications, the ability of attribute a result to a reasoning process from the network is in high demand for robust scientific generalizations to hold. In this work we aim to motivate the need for and demonstrate the use of post-hoc explainability methods when applied to AI methods used in scientific applications. To this end, we introduce explainability add-ons to the existing graph neural network (GNN) for neutrino tagging, NuGraph2. The explanations take the form of a suite of techniques examining the output of the network (node classifications) and the edge connections between them, and probing of the latent space using novel general-purpose tools applied to this network. We show how none of these methods are singularly sufficient to show network ‘understanding’, but together can give insights into the processes used in classification. While these methods are tested on the NuGraph2 application, they can be applied to a broad range of networks, not limited to GNNs. The code for this work is publicly available on GitHub at https://github.com/voetberg/XNuGraph .
Abstract Modeling temporal dependencies within quantum systems remains a key challenge for quantum machine learning. Current quantum neural networks largely depend on classical recurrent modules, which introduce optimization bottlenecks and coherence loss during sequence processi…
Abstract The lottery ticket hypothesis (LTH) posits that within overparametrized neural networks, there exist sparse subnetworks that are capable of matching the performance of the original model when trained in isolation from the original initialization. We extend this hypothesi…
Abstract Machine learning is increasingly utilized in molecular dynamics simulations to investigate complex system properties across disciplines ranging from chemical and physical sciences to engineering. However, these methods often require large datasets for model training and…
Abstract Nuclear energy is a clean, reliable power source, but realizing its potential requires strict safety measures in nuclear power plants. Thermal-hydraulic (TH) codes are used to simulate potential accident scenarios in probabilistic safety assessment (PSA). Their high comp…
Abstract Tuning of gate-defined semiconductor quantum dots (QDs) is a major bottleneck for scaling spin-qubit technologies. We present a deep learning driven, semantic-segmentation pipeline that performs charge auto-tuning by locating transition lines in full charge stability dia…
Abstract Physics-informed neural networks (PINNs) on complex domains are limited by input representations that encode geometry but do not reflect the physics of the governing PDE. We propose an operator-aware PINN for solid mechanics problems that embeds precomputed eigenmodes of…