arXiv subjects

Marianna La Rocca

Publications and source records attributed to Marianna La Rocca.

7 recordsLinked to original sources

arXiv2026

Data geometry preserves prediction but reshapes explanation

Across scientific domains, empirical data often give rise to positive semidefinite matrices that encode similarities, couplings or interactions and induce natural geometries. We investigate whether changing the geometry used to compare the same empirical representations preserves subject-level performance and the local structures from which explanations are derived, a question we term the prediction explanation invariance problem. We address this problem across morphometric MRI, fMRI and EEG data by comparing Frobenius and trace geometries applied to the same empirical matrices. Predictive performance was broadly preserved across geometries, but comparable performance did not imply agreement in subject-level decision scores or predicted labels. Predictive similarity also masked geometry-dependent differences in local neighbourhoods, perturbation sensitivities and explanatory rankings. Dimensionality reduction made the two subject-space geometries progressively more concordant, while classification performance and decision-level agreement declined. These results identify data geometry as a hidden degree of freedom in explainability: similar performance does not guarantee invariant decisions or explanations.

physics.data-an↗

arXiv2026

A Tensor Network Framework for Interpretable Graph Analysis of Brain Networks

Identifying robust neurobiological signatures of brain disorders requires machine learning approaches that combine predictive performance with interpretable representations of feature interactions. Here we introduce a quantum-inspired framework based on tensor network machine learning that learns distributed representations of gray-matter features encoded in a Matrix Product State representation, a variational ansatz originally developed for quantum many-body systems. The trained model is then used not only as a classifier but to extract quantum connected correlations between features, which encode higher-order feature interactions and which we map onto a weighted graph. This construction allows us to track, within a single representation, both: (i) the global spectral properties of the network (capturing collective learning dynamics), and (ii) node-level centrality measures (providing interpretable signatures of individual brain regions). Using repeated train-test sampling schemes, we analyze two classification tasks on structural MRI data as examples of complex brain disorders: healthy controls versus schizophrenia and versus bipolar disorder. Node-level analysis identifies a stable set of gray-matter features, most prominently Heschl gyrus, insular cortex, and frontal regions, that act as hubs across multiple centrality measures and across resamplings. These centralities display lower variability across resamplings than Shapley values, supporting the interpretive value of the network representation. The bipolar feature set emerges as a subset of the schizophrenia one, consistent with the hierarchically organized neuroanatomical alterations reported in neuroimaging studies and offering a network-based characterization of this hierarchy.

q-bio.NC↗

arXiv2025

Transfer entropy and O-information to detect grokking in tensor network multi-class classification problems

Quantum-enhanced machine learning, encompassing both quantum algorithms and quantum-inspired classical methods such as tensor networks, offers promising tools for extracting structure from complex, high-dimensional data. In this work, we study the training dynamics of Matrix Product State (MPS) classifiers applied to three-class problems, using both fashion MNIST and hyper-spectral satellite imagery as representative datasets. We investigate the phenomenon of grokking, where generalization emerges suddenly after memorization, by tracking entanglement entropy, local magnetization, and model performance across training sweeps. Additionally, we employ information theory tools to gain deeper insights: transfer entropy is used to reveal causal dependencies between label-specific quantum masks, while O-information captures the shift from synergistic to redundant correlations among class outputs. Our results show that grokking in the fashion MNIST task coincides with a sharp entanglement transition and a peak in redundant information, whereas the overfitted hyper-spectral model retains synergistic, disordered behavior. These findings highlight the relevance of high-order information dynamics in quantum-inspired learning and emphasize the distinct learning behaviors that emerge in multi-class classification, offering a principled framework to interpret generalization in quantum machine learning architectures.

quant-ph↗

arXiv2025

Grokking as an entanglement transition in tensor network machine learning

Grokking is a intriguing phenomenon in machine learning where a neural network, after many training iterations with negligible improvement in generalization, suddenly achieves high accuracy on unseen data. By working in the quantum-inspired machine learning framework based on tensor networks, we numerically prove that grokking phenomenon can be related to an entanglement dynamical transition in the underlying quantum many-body systems, consisting in a one-dimensional lattice with each site hosting a qubit. Two datasets are considered as use case scenarios, namely fashion MNIST and gene expression communities of hepatocellular carcinoma. In both cases, we train Matrix Product State (MPS) to perform binary classification tasks, and we analyse the learning dynamics. We exploit measurement of qubits magnetization and correlation functions in the MPS network as a tool to identify meaningful and relevant gene subcommunities, verified by means of enrichment procedures.

quant-ph↗

arXiv2021

Efficient and Visualizable Convolutional Neural Networks for COVID-19 Classification Using Chest CT

With COVID-19 cases rising rapidly, deep learning has emerged as a promising diagnosis technique. However, identifying the most accurate models to characterize COVID-19 patients is challenging because comparing results obtained with different types of data and acquisition processes is non-trivial. In this paper we designed, evaluated, and compared the performance of 20 convolutional neutral networks in classifying patients as COVID-19 positive, healthy, or suffering from other pulmonary lung infections based on Chest CT scans, serving as the first to consider the EfficientNet family for COVID-19 diagnosis and employ intermediate activation maps for visualizing model performance. All models are trained and evaluated in Python using 4173 Chest CT images from the dataset entitled "A COVID multiclass dataset of CT scans," with 2168, 758, and 1247 images of patients that are COVID-19 positive, healthy, or suffering from other pulmonary infections, respectively. EfficientNet-B5 was identified as the best model with an F1 score of 0.9769+/-0.0046, accuracy of 0.9759+/-0.0048, sensitivity of 0.9788+/-0.0055, specificity of 0.9730+/-0.0057, and precision of 0.9751 +/- 0.0051. On an alternate 2-class dataset, EfficientNetB5 obtained an accuracy of 0.9845+/-0.0109, F1 score of 0.9599+/-0.0251, sensitivity of 0.9682+/-0.0099, specificity of 0.9883+/-0.0150, and precision of 0.9526 +/- 0.0523. Intermediate activation maps and Gradient-weighted Class Activation Mappings offered human-interpretable evidence of the model's perception of ground-class opacities and consolidations, hinting towards a promising use-case of artificial intelligence-assisted radiology tools. With a prediction speed of under 0.1 seconds on GPUs and 0.5 seconds on CPUs, our proposed model offers a rapid, scalable, and accurate diagnostic for COVID-19.

eess.IV↗

arXiv2020

Prediction of Epilepsy Development in Traumatic Brain Injury Patients from Diffusion Weighted MRI

Post-traumatic epilepsy (PTE) is a life-long complication of traumatic brain injury (TBI) and is a major public health problem that has an estimated incidence that ranges from 2%-50%, depending on the severity of the TBI. Currently, the pathomechanism that in-duces epileptogenesis in TBI patients is unclear, and one of the most challenging goals in the epilepsy community is to predict which TBI patients will develop epilepsy. In this work, we used diffusion-weighted imaging (DWI) of 14 TBI patients recruited in the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy (EpiBioS4Rx)to measure and analyze fractional anisotropy (FA), obtained from tract-based spatial statistic (TBSS) analysis. Then we used these measurements to train two support vector machine (SVM) models to predict which TBI patients have developed epilepsy. Our approach, tested on these 14 patients with a leave-two-out cross-validation, allowed us to obtain an accuracy of 0.857 $\pm$ 0.18 (with a 95% level of confidence), demonstrating it to be potentially promising for the early characterization of PTE.

q-bio.QM↗

arXiv2017

Brain structural connectivity atrophy in Alzheimer's disease

Analysis and quantification of brain structural changes, using Magnetic resonance imaging (MRI), are increasingly used to define novel biomarkers of brain pathologies, such as Alzheimer's disease (AD). Network-based models of the brain have shown that both local and global topological properties can reveal patterns of disease propagation. On the other hand, intra-subject descriptions cannot exploit the whole information context, accessible through inter-subject comparisons. To address this, we developed a novel approach, which models brain structural connectivity atrophy with a multiplex network and summarizes it within a classification score. On an independent dataset multiplex networks were able to correctly segregate, from normal controls (NC), AD patients and subjects with mild cognitive impairment that will convert to AD (cMCI) with an accuracy of, respectively, $0.86 \pm 0.01$ and $0.84 \pm 0.01$. The model also shows that illness effects are maximally detected by parceling the brain in equal volumes of $3000$ $mm^3$ ("patches"), without any $a$ $priori$ segmentation based on anatomical features. A direct comparison to standard voxel-based morphometry on the same dataset showed that the multiplex network approach had higher sensitivity. This method is general and can have twofold potential applications: providing a reliable tool for clinical trials and a disease signature of neurodegenerative pathologies.

physics.med-ph↗