arXiv subjects

Davide D'Ascenzo

Publications and source records attributed to Davide D'Ascenzo.

7 recordsLinked to original sources

arXiv2026

Efficient and Scalable Provenance Tracking for LLM-Generated Code Snippets

Large language models (LLMs) for code completion and generation are increasingly used in software development, yet they may reproduce training examples verbatim and without authorship attribution, raising legal and ethical concerns around plagiarism and license compliance. Classical fingerprint-based plagiarism detectors, such as Winnowing, remain highly effective, yet the inspection requires comparing fragments of code to the entire training set, and their linear-time search makes them impractical for the billion-scale corpora used to train modern code LLMs. To bridge this gap, we introduce SourceTracker, a 300M-parameter encoder tailored for code retrieval, together with a hybrid two-stage provenance-tracking pipeline HybridSourceTracker (HST). HST first narrows down a small set of candidate snippets via vector search, then re-ranks those candidates using Winnowing on exact fingerprints. We train and evaluate our system on a 10M-snippet subset of the TheStackV2 dataset, with both verbatim and adapted snippets that emulate realistic identifier renaming. On an in vitro 100k-snippet search space with adapted queries, our hybrid approach reaches a mean reciprocal rank on par with Winnowing for 30-token fragments. Then, starting from windows >= 60 tokens, it consistently overperforms by up to 5.4% while preserving logarithmic-time query complexity. In a complementary evaluation using an LLM-based judge, we find that many retrieved snippets not labeled as ground truth are still highly similar to the expected sources, particularly with longer context windows, and thus remain useful for end users. Overall, our results demonstrate that integrating vector search with fingerprinting enables scalable, high precision provenance tracking for code produced by LLMs, provided that their training data is made accessible...

cs.SE↗

arXiv2026

Semi-Monotonicity for Spectral Centrality Measures

Score monotonicity and rank monotonicity are properties describing the behavior of a centrality measure when an arc is added to a network: the former requires that the score of the target of the arc should increase, the latter that its importance with respect to the remaining nodes should not deteriorate. While in directed networks almost all classical centrality measures satisfy both properties, in undirected networks they fail for most measures: adding an edge can reduce the score or the rank of one of its endpoints. Semi-monotonicity is a recently introduced weaker property for undirected networks, requiring that at least one of the two endpoints of the new edge enjoys monotonicity, and it is known to hold for closeness, harmonic centrality, distance-decay centralities and betweenness. In this paper we study semi-monotonicity for three classical spectral centrality measures: eigenvector centrality, Katz's index, and PageRank. We prove that all of them are strictly rank semi-monotone on connected undirected graphs, and that all of them are also score semi-monotone, (the only exception being eigenvector centrality when scores are normalized by projecting the constant vector on the dominant eigenspace, for which we provide a counterexample). In particular, adding an edge can never decrease the PageRank of both its endpoints, which answers a question left open in previous work.

cs.SI↗

arXiv2026

Score Distributions, Not Cells: Evaluating Single-Cell Perturbations Under Class Overlap

Most classification problems assume the classes are roughly separable, so that an individual sample can usually be assigned to one class. Single-cell perturbation data violates this assumption: two perturbations can produce different populations of cells while overlapping so much that an individual cell could belong to either. Per-cell accuracy then measures this overlap rather than model quality. We see this on Tahoe-100M and the Virtual Cell Challenge, where a linear classifier, an MLP, and a Transformer all plateau near macro-F1 0.2-0.3 even though almost every pair of perturbations is statistically distinguishable. The fix is to score perturbations across the whole population rather than cell by cell. We average a classifier's per-cell probability vectors over all cells of a perturbation to form a population profile, then rank candidate perturbations by this profile; we call the resulting score the Classifier Discrimination Score (CDS). Taking the top-ranked class recovers the winning perturbation. It needs no retraining, costs linear time in the number of cells, and recovers near-perfect identification from the same weak models. CDS differs from the pseudobulk-based Perturbation Discrimination Score (PDS) used in recent benchmarks only in where the average is taken, raw gene expression for PDS versus a learned discriminative space for CDS, and identifies the true perturbation more reliably on both datasets, with the gap widening as cells grow scarce. Because a metric that misranks the ground truth will misrank the models scored against it, per-cell accuracy and raw-pseudobulk scores should be used with caution when comparing perturbation models.

cs.LG↗

arXiv2026

scDataset: Scalable Data Loading for Deep Learning on Large-Scale Single-Cell Omics

Training deep learning models on single-cell datasets with hundreds of millions of cells requires loading data from disk, as these datasets exceed available memory. While random sampling provides the data diversity needed for effective training, it is prohibitively slow due to the random access pattern overhead, whereas sequential streaming achieves high throughput but introduces biases that degrade model performance. We present scDataset, a PyTorch data loader that enables efficient training from on-disk data with seamless integration across diverse storage formats. Our approach combines block sampling and batched fetching to achieve quasi-random sampling that balances I/O efficiency with minibatch diversity. On Tahoe-100M, a dataset of 100 million cells, scDataset achieves more than two orders of magnitude speedup compared to true random sampling while working directly with AnnData files. We provide theoretical bounds on minibatch diversity and empirically show that scDataset matches the performance of true random sampling across multiple classification tasks and model architectures.

cs.LG↗

arXiv2026

When Labels Have Structure: Improving Image Classification with Hierarchy-Aware Cross-Entropy

Standard cross-entropy is the default classification loss across virtually all of machine learning, yet it treats all misclassifications equally, ignoring the semantic distances that a class hierarchy encodes. We propose Hierarchy-Aware Cross-Entropy (HACE), a drop-in replacement for standard cross-entropy that incorporates a known class hierarchy directly into the loss. HACE combines two components: prediction aggregation, which propagates the model's probability mass upward through the class hierarchy to ensure that parent nodes accumulate the confidence of their children; and ancestral label smoothing, which distributes the ground-truth signal along the path from the true class to the root. We evaluate HACE on CIFAR-100, FGVC Aircraft, and NABirds in two regimes: end-to-end training across six architectures spanning convolutional and attention-based designs, and linear probing on frozen DINOv2-Large features. In end-to-end training, HACE improves accuracy over standard cross-entropy in 15 out of 18 architecture--dataset pairs, with a mean gain of 4.66\%. In linear probing on frozen DINOv2-Large features, HACE outperforms all competing methods on all three datasets, with a mean improvement of 2.18\% over the next best baseline.

cs.LG↗

arXiv2025

Position: A Theory of Deep Learning Must Include Compositional Sparsity

Overparametrized Deep Neural Networks (DNNs) have demonstrated remarkable success in a wide variety of domains too high-dimensional for classical shallow networks subject to the curse of dimensionality. However, open questions about fundamental principles, that govern the learning dynamics of DNNs, remain. In this position paper we argue that it is the ability of DNNs to exploit the compositionally sparse structure of the target function driving their success. As such, DNNs can leverage the property that most practically relevant functions can be composed from a small set of constituent functions, each of which relies only on a low-dimensional subset of all inputs. We show that this property is shared by all efficiently Turing-computable functions and is therefore highly likely present in all current learning problems. While some promising theoretical insights on questions concerned with approximation and generalization exist in the setting of compositionally sparse functions, several important questions on the learnability and optimization of DNNs remain. Completing the picture of the role of compositional sparsity in deep learning is essential to a comprehensive theory of artificial, and even general, intelligence.

cs.LG↗

arXiv2023

Score and Rank Semi-Monotonicity for Closeness, Betweenness and Harmonic Centrality

In the study of the behavior of centrality measures with respect to network modifications, score monotonicity means that adding an arc increases the centrality score of the target of the arc; rank monotonicity means that adding an arc improves the importance of the target of the arc relative to the remaining nodes. It is known that score and rank monotonicity hold in directed graphs for almost all centrality measures. In undirected graphs one expects that the corresponding properties (where both endpoints of the new edge enjoy the increase in score/rank) hold when adding a new edge. However, recent results have shown that in undirected networks this is not true: for many centrality measures, it is possible to find situations where adding an edge reduces the rank of one of its two endpoints. In this paper we introduce a weaker condition for undirected networks, semi-monotonicity, in which just one of the endpoints of a new edge is required to enjoy score or rank monotonicity. We show that this condition is satisfied by closeness and betweenness centrality, and that harmonic centrality satisfies it in an even stronger sense.

cs.SI↗