How Architecture and Training Affect TPC Representations Across Experiments

Tyler Wheeler; Michelle P. Kuchera; Raghuram Ramanujan; William Sieland; Ryan Krupp; Yassid Ayyad; Daniel Bazin; Connor L. Cross

cs.LGAug 22, 2026

How Architecture and Training Affect TPC Representations Across Experiments

Authors: Tyler Wheeler, Michelle P. Kuchera, Raghuram Ramanujan, William Sieland, Ryan Krupp, Yassid Ayyad, Daniel Bazin, Connor L. Cross,

Organizations: Department of Computational Math, Science, and Engineering, Michigan State University, East Lansing, MI, USA · Department of Physics, Grand Valley State University, Allendale, Michigan, USA · Department of Physics, Davidson College, Davidson, NC, USA · Department of Mathematics and Computer Science, Davidson College, Davidson, NC, USA ·

Abstract

Deep-learning efforts have increasingly shifted toward foundation model approaches. In experimental physics, this allows models and learned representations to be reused beyond the experiments in which they were developed. This work evaluates the reusability of representations across experiments and detector systems using probes on frozen encoders. These probes reveal task-relevant structure before downstream adaptation, complementing fine-tuning. Together with random-weight controls, they distinguish contributions from architecture and encoder training that downstream performance alone cannot resolve. Time projection chamber (TPC) data provide a useful testbed because events from TPC systems can be represented as variable-length sparse tensors, while detector geometries, event topologies, and scientific tasks can differ substantially. We investigate whether fixed-dimensional TPC event representations can be reused across classification tasks, experiments, and detector systems. Sparse ResNet and PointNet-style encoders produce 512-dimensional embeddings for four datasets from the GADGET II TPC and AT-TPC. Randomly initialized encoders isolate the contribution from architecture before supervised training. We then train each encoder on a classification task, freeze its parameters, and train a linear or nonlinear probe for each downstream task. We find that this architecture-induced structure remains useful across experiments and detector systems. The randomly initialized PointNet-style representation is highly informative on several tasks. The two architectures organize their embedding spaces differently, but neither exhibits a large, systematic loss of utility cross-detector. These results show that architecture is a major source of task-relevant structure in TPC embeddings and should be treated explicitly when assessing representation learning and developing reusable detector models.

Figures & tables

Panel of Figure 1 — Figure 1: Example event topologies from the two TPC systems used in this work. (a) A 20 Mg, 1600 keV beta-delayed proton event from the GADGET II TPC. (b) A 4-track event from the AT-TPC.

Panel of Figure 2 — Figure 2: (a) Schematic of the GAseous Detector with GErmanium Tagging (GADGET) II TPC—a detector optimized for measuring low-energy $\beta$ -delayed particle decays, surrounded by a germanium array. (b) Schematic of the Active-Target TPC (AT-TPC)—a detector designed for nuclear reaction studies in inverse kinematics operated in a solenoid magnet providing up to 4 T of magnetic field.

Feature	GADGET II TPC	AT-TPC
Primary use case	$\beta$ -delayed charged-particle decays	Inverse-kinematics reactions
Detector mode	Implant-decay TPC	Active-target TPC
Active/drift length	40 cm drift region	100 cm drift region
Active diameter	8 cm pad-plane diameter	55 cm pad-plane diameter
Gas/target role	Stopping/decay medium	Reaction target and tracking medium
Typical gas	P10 at 800 Torr	Experiment-dependent target gas (e.g., 4 He, 2 H)

Table 1: Comparison of the two TPC systems used in this work.

Dataset	Detector	Isotope	Task name	Classes	Events per class	Total events
E21072	GADGET II	20 Mg	20 Mg, 2 class	Proton; alpha	25,078; 9,501	34,579
E21072	GADGET II	20 Mg	20 Mg, 3 class	770 keV p; 1596 keV p; 2153 keV $\alpha$	18,770; 6,308; 9,501	34,579
E25058	GADGET II	21 Mg	21 Mg, 3 class	1285 keV p; 1930 keV p; 2153 keV $\alpha$	1,261; 1,351; 3,363	5,975
E20020	AT-TPC	16 O	16 O $+\alpha$ , 2 class	0–2 tracks; 3–5 tracks	1,184; 1,274	2,458
E20020	AT-TPC	16 O	16 O $+\alpha$ , 3 class	0–2 tracks; 3 tracks; 4–5 tracks	1,184; 476; 798	2,458
E20009	AT-TPC	10 B	10 B $+d$ , 2 class	3 tracks; 4 tracks	2,310; 2,151	4,461

Table 2: Summary of the datasets used in this work.

Figure 3: Sparse ResNet14 architecture used to produce event-level TPC embeddings. The lower panel shows the structure of each residual block.

Figure 4: Minkowski PointNet-style architecture used to embed TPC events. Top: a shared multilayer perceptron is applied independently to each of the input points. Global max pooling aggregates the resulting

N\times 512

features into the 512-dimensional event embedding. The post-pooling linear layer, dropout, and final classifier are used during source-task training but are not included in the extracted embedding. Bottom: each pointwise layer applies the same learned transformation independently to every active input; information is combined across inputs only by global pooling.

Evaluation task	Encoder training	Architecture	Linear SVM		RBF SVM
			Accuracy	Macro-F1	Accuracy	Macro-F1
${}^{20}\mathrm{Mg}$ , 3 class	Naïve baseline	—	$0.5428\pm 0.0000$	$0.2346\pm 0.0000$	$0.5428\pm 0.0000$	$0.2346\pm 0.0000$
	Random weights	ResNet	$0.9109\pm 0.0052$	$0.8840\pm 0.0068$	$0.9321\pm 0.0061$	$0.9077\pm 0.0091$
	Random weights	PointNet	$0.9979\pm 0.0005$	$0.9977\pm 0.0004$	$0.9933\pm 0.0013$	$0.9928\pm 0.0015$
	${}^{20}\mathrm{Mg}$ , 2 class	ResNet	$0.9782\pm 0.0059$	$0.9732\pm 0.0073$	$0.9803\pm 0.0042$	$0.9753\pm 0.0053$
	${}^{20}\mathrm{Mg}$ , 2 class	PointNet	$\mathbf{0.9999\pm 0.0001}$	$\mathbf{0.9999\pm 0.0001}$	$\mathbf{0.9999\pm 0.0001}$	$\mathbf{0.9999\pm 0.0001}$

Table 3: Within-experiment probe performance. Values are the mean

\pm

standard deviation over five runs. Boldface marks the largest mean in each evaluation-task and metric column. The naïve baseline predicts the most frequent class in the training set.

Evaluation task	Encoder training	Architecture	Linear SVM		RBF SVM
			Accuracy	Macro-F1	Accuracy	Macro-F1
${}^{21}\mathrm{Mg}$ , 3 class	Naïve baseline	—	$0.5632\pm 0.0000$	$0.2402\pm 0.0000$	$0.5632\pm 0.0000$	$0.2402\pm 0.0000$
	Random weights	ResNet	$0.8524\pm 0.0051$	$0.7985\pm 0.0065$	$0.8499\pm 0.0142$	$0.7922\pm 0.0221$
	Random weights	PointNet	$0.9955\pm 0.0025$	$0.9942\pm 0.0031$	$0.9945\pm 0.0027$	$0.9937\pm 0.0030$
	${}^{20}\mathrm{Mg}$ , 2 class	ResNet	$0.9506\pm 0.0026$	$0.9246\pm 0.0042$	$0.9622\pm 0.0065$	$0.9422\pm 0.0098$
	${}^{20}\mathrm{Mg}$ , 2 class	PointNet	$\mathbf{0.9967\pm 0.0032}$	$\mathbf{0.9958\pm 0.0039}$	$\mathbf{0.9985\pm 0.0006}$	$\mathbf{0.9983\pm 0.0008}$

Table 4: Cross-experiment probe performance. Values are the mean

\pm

standard deviation over five runs. Boldface marks the largest mean in each evaluation-task and metric column. The naïve baseline predicts the most frequent class in the training set.

Evaluation task	Encoder training	Architecture	Linear SVM		RBF SVM
			Accuracy	Macro-F1	Accuracy	Macro-F1
${}^{16}\mathrm{O}+\alpha$ , 3 class	Naïve baseline	—	$0.4817\pm 0.0000$	$0.2167\pm 0.0000$	$0.4817\pm 0.0000$	$0.2167\pm 0.0000$
	Random weights	ResNet	$0.6008\pm 0.0081$	$0.5107\pm 0.0107$	$0.6646\pm 0.0121$	$0.4873\pm 0.0070$
	Random weights	PointNet	$0.8008\pm 0.0173$	$0.7579\pm 0.0197$	$0.8390\pm 0.0081$	$0.8044\pm 0.0101$
	${}^{20}\mathrm{Mg}$ , 2 class	ResNet	$0.6955\pm 0.0083$	$0.6162\pm 0.0060$	$0.7817\pm 0.0068$	$0.6960\pm 0.0172$
	${}^{20}\mathrm{Mg}$ , 2 class	PointNet	$\mathbf{0.8248\pm 0.0115}$	$\mathbf{0.7841\pm 0.0175}$	$\mathbf{0.8720\pm 0.0110}$	$\mathbf{0.8489\pm 0.0135}$

Table 5: Cross-detector probe performance. Values are the mean

\pm

standard deviation over five runs. Boldface marks the largest mean in each evaluation-task and metric column. The naïve baseline predicts the most frequent class in the training set.

Figure 5: PCA projections of the latent representations for the ${}^{16}\mathrm{O}+\alpha$ three-class task. Panels compare randomly initialized and 16 O-trained ResNet and PointNet-style encoders. Points are colored according to track-multiplicity class.

Figure 6: PCA projections of the latent representations for the 21 Mg three-class task. Panels compare randomly initialized and 16 O-trained ResNet and PointNet-style encoders. Points are colored according to the three decay-event classes.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Setting	ResNet14	PointNet-style
Loss	Cross-entropy	Cross-entropy
Optimizer	AdamW	AdamW
Initial learning rate	$5\times 10^{-4}$	$5\times 10^{-4}$ ( ${}^{16}\mathrm{O}+\alpha$ ); $5\times 10^{-6}$ ( ${}^{20}\mathrm{Mg}$ )
Weight decay	$10^{-4}$	$10^{-4}$
Batch size	128	128
Learning-rate scheduler	Reduce on plateau	Reduce on plateau

Appendix

Table 6: Optimization, regularization, and augmentation settings used to train the ResNet14 and PointNet-style encoders. Settings were used across encoder-training datasets unless otherwise indicated.

Representation Learning

Representation learning develops data transformations that expose useful structure for prediction, transfer, generation, and control.

See all 804 papers

Beyond Task Success: A Theory of Representation Reuse
Oct 13, 2026Zibo Meng, Andrew C. M. Austin, Alin AchimRepresentation Learning Transfer Learning
Cheap Probes Predict Expensive Training in 3D-CT Vision--Language Models
Jul 24, 2026Renjie LiangVLM Evaluation Medical VLMs
University of Florida, Gainesville, FL, USA
TAP-Path: Task-Adaptive Structural and Token Pruning for Efficient and Trustworthy Pathology Foundation Models
Sep 3, 2026Mehedi Hasan, Ashfak Yeafi, Md Khairul IslamComputational Pathology Pathology Foundation Models
Department of Computer Science and Engineering, Brac University, Dhaka, Bangladesh · EEE, Khulna University of Engineering & Technology, Khulna, Bangladesh · Mathematics and Computer Science, Hobart and William Smith Colleges, Geneva, NY, USA

Oct 13, 2026J/K move · Enter open · S save

Beyond Task Success: A Theory of Representation Reuse

Zibo Meng, Andrew C. M. Austin, Alin Achim

Representation Learning Transfer Learning

Open paper AskPDF arXiv

A learned representation is useful beyond its source task only if later tasks can make use of what the encoder has already learned. Source-task performance, however, gives no guarantee that the representation itself will remain useful once the original head is removed. We study representation reuse with respect to a specified family of future tasks. For such a family, we define the task-relevant state as the coarsest deterministic state sufficient for all of its tasks. A larger task family can require a finer state, and losing a required distinction makes at least one future task unrecoverable on a target support containing that conflict. Exact reuse also depends on how the retained information is presented to the downstream rule. We separate this into structural accessibility, which concerns whether the required distinctions remain recoverable, and coordinate consistency, which concerns whether those distinctions can be interpreted by the same downstream rule across samples. Both failure modes appear separately in the experiments. Reuse improves as task-relevant distinctions become more accessible. Reversible sample-dependent coordinate changes can sharply reduce reuse even when the representation remains recoverable, and inverting the applicable coordinate change restores performance. The same accessibility-reuse ordering appears in prospectively specified CIFAR-100 experiments with both ResNet-18 and DenseNet-121. Finally, preserving a candidate state that contains the distinctions needed by the anticipated task family improves reuse on compatible held-out tasks, with the clearest benefit when target supervision is limited. Representation reuse is therefore relative to the tasks that follow source learning: a useful representation must retain the distinctions those tasks need and make them available in a form that a shared downstream rule can use.

Abstract

Representation Learning

Explore similar work

Beyond Task Success: A Theory of Representation Reuse

Cheap Probes Predict Expensive Training in 3D-CT Vision--Language Models

TAP-Path: Task-Adaptive Structural and Token Pruning for Efficient and Trustworthy Pathology Foundation Models