Finding and using interpretable latents in a neutrino foundation model with sparse autoencoders

• URL: http://arxiv.org/abs/2608.26090v1
• Date: Wed, 26 Aug 2026 17:53:00 GMT
• Title: Finding and using interpretable latents in a neutrino foundation model with sparse autoencoders
• Abstract summary: We present a first application of sparse-autoencoder-based mechanistic interpretability to particle physics.<n>We study a neutrino foundation model pretrained on IceCube data and fine-tuned for direction reconstruction.<n>Motivated by this underused information, we train an uncertainty head on the same event-level representation to predict the model's angular reconstruction error.
• Score: 3.3461260868741527
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We present a first application of sparse-autoencoder-based mechanistic interpretability to particle physics. Studying a neutrino foundation model pretrained on IceCube data and fine-tuned for direction reconstruction, we identify a validated atlas of physical concepts in the model representation, using a strict validation protocol consisting of held-out tests, matched nuisance controls, and replication across independent dictionary trainings. Causal interventions show that the direction head barely draws on this atlas. Motivated by this underused information, we train an uncertainty head on the same event-level representation to predict the model's angular reconstruction error. Unlike the direction head, it depends causally on quality and brightness features from the atlas. At $20\%$ selection efficiency, this interpretable estimator improves the median angular resolution from $20.2^\circ$ to $3.2^\circ$. These results suggest that mechanistic interpretability can reveal learned latent physics encoded within a model's internal representation and help design downstream tasks that exploit it.

Related papers

• Dissecting Jet-Tagger Through Mechanistic Interpretability [0.0]
  We perform a mechanistic interpretability analysis of the Particle Transformer architecture, trained on the Top Quark Tagging reference dataset.<n>We identify the computational circuit responsible for jet classification and characterizing the physical content of its internal representations.
  arXiv  Detail & Related papers  (2026-05-11T02:11:47Z)
• Decoding the Critique Mechanism in Large Reasoning Models [50.821607345799386]
  Large Reasoning Models (LRMs) exhibit backtracking and self-verification mechanisms that enable them to revise intermediate steps and reach correct solutions.<n>We investigate how current LRMs recover from errors by inserting arithmetic mistakes in their intermediate reasoning steps.<n>We find that despite the error propagating through the chain-of-thought, the model still reaches the correct final answer.
  arXiv  Detail & Related papers  (2026-03-17T10:03:30Z)
• Attribution-Guided Model Rectification of Unreliable Neural Network Behaviors [60.06461883533697]
  We leverage rank-one model editing to establish an attribution-guided model rectification framework.<n>We first distinguish our rectification setting from existing model editing, yielding a formulation that corrects unreliable behavior.<n>To target the primary source of misbehavior, we introduce an attribution-guided layer localization method.
  arXiv  Detail & Related papers  (2026-03-08T01:06:07Z)
• SINDER: Repairing the Singular Defects of DINOv2 [61.98878352956125]
  Vision Transformer models trained on large-scale datasets often exhibit artifacts in the patch token they extract. We propose a novel fine-tuning smooth regularization that rectifies structural deficiencies using only a small dataset.
  arXiv  Detail & Related papers  (2024-07-23T20:34:23Z)
• Learning Active Subspaces and Discovering Important Features with Gaussian Radial Basis Functions Neural Networks [0.0]
  We show that precious information is contained in the spectrum of the precision matrix that can be extracted once the training of the model is completed. We conducted numerical experiments for regression, classification, and feature selection tasks. Our results demonstrate that the proposed model does not only yield an attractive prediction performance compared to the competitors.
  arXiv  Detail & Related papers  (2023-07-11T09:54:30Z)
• Disentanglement via Latent Quantization [60.37109712033694]
  In this work, we construct an inductive bias towards encoding to and decoding from an organized latent space. We demonstrate the broad applicability of this approach by adding it to both basic data-re (vanilla autoencoder) and latent-reconstructing (InfoGAN) generative models.
  arXiv  Detail & Related papers  (2023-05-28T06:30:29Z)
• Leveraging Angular Information Between Feature and Classifier for Long-tailed Learning: A Prediction Reformulation Approach [90.77858044524544]
  We reformulate the recognition probabilities through included angles without re-balancing the classifier weights. Inspired by the performance improvement of the predictive form reformulation, we explore the different properties of this angular prediction. Our method is able to obtain the best performance among peer methods without pretraining on CIFAR10/100-LT and ImageNet-LT.
  arXiv  Detail & Related papers  (2022-12-03T07:52:48Z)
• Recoding latent sentence representations -- Dynamic gradient-based activation modification in RNNs [0.0]
  In RNNs, encoding information in a suboptimal way can impact the quality of representations based on later elements in the sequence. I propose an augmentation to standard RNNs in form of a gradient-based correction mechanism. I conduct different experiments in the context of language modeling, where the impact of using such a mechanism is examined in detail.
  arXiv  Detail & Related papers  (2021-01-03T17:54:17Z)
• Autoencoding Variational Autoencoder [56.05008520271406]
  We study the implications of this behaviour on the learned representations and also the consequences of fixing it by introducing a notion of self consistency. We show that encoders trained with our self-consistency approach lead to representations that are robust (insensitive) to perturbations in the input introduced by adversarial attacks.
  arXiv  Detail & Related papers  (2020-12-07T14:16:14Z)
• Supervised Learning in the Presence of Concept Drift: A modelling framework [5.22609266390809]
  We present a modelling framework for the investigation of supervised learning in non-stationary environments. We model two example types of learning systems: prototype-based Learning Vector Quantization (LVQ) for classification and shallow, layered neural networks for regression tasks.
  arXiv  Detail & Related papers  (2020-05-21T09:13:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.