arXiv subjects

Riccardo Cantoro

Publications and source records attributed to Riccardo Cantoro.

3 recordsLinked to original sources

arXiv2026

GRIP: Geometric Refinement and Adaptive Information Potential for Data Efficiency

The performance of Large Language Models (LLMs) is increasingly governed by data efficiency rather than raw scaling volume. However, existing selection methods often decouple global distribution balancing from local instance selection, compromising the hierarchical integrity of the training set. We introduce \textbf{GRIP} (Geometric Refinement and Adaptive Information Potential), a framework that unifies these dimensions by modeling the corpus as an information-dense geometric space. GRIP employs a \textbf{Rapid Adaptation Probe (RAP)} to quantify the information potential of semantic clusters, dynamically re-allocating the sampling budget to regions with the highest representation deficits. Subsequently, we perform Intra-Cluster Selection using a \textbf{length-rectified geometric prior} to counteract embedding density artifacts and preserve long-tail logical sequences. Extensive evaluations on Mixture-of-Experts (MoE) models up to 300B tokens demonstrate that GRIP consistently outperforms state-of-the-art baselines, \textbf{surpassing the performance of models trained on $3\times$ larger uncurated datasets}. Our work establishes a robust geometric foundation for adaptive data curation in large-scale pre-training.

cs.CL↗

arXiv2026

UniGeM: Unifying Data Mixing and Selection via Geometric Exploration and Mining

The scaling of Large Language Models (LLMs) is increasingly limited by data quality. Most methods handle data mixing and sample selection separately, which can break the structure in code corpora. We introduce \textbf{UniGeM}, a framework that unifies mixing and selection by treating data curation as a \textit{manifold approximation} problem without training proxy models or relying on external reference datasets. UniGeM operates hierarchically: \textbf{Macro-Exploration} learns mixing weights with stability-based clustering; \textbf{Micro-Mining} filters high-quality instances by their geometric distribution to ensure logical consistency. Validated by training 8B and 16B MoE models on 100B tokens, UniGeM achieves \textbf{2.0$\times$ data efficiency} over a random baseline and further improves overall performance compared to SOTA methods in reasoning-heavy evaluations and multilingual generalization.

cs.LG↗

arXiv2024

COVID-19 Detection from Exhaled Breath

The SARS-CoV-2 coronavirus emerged in 2019, causing a COVID-19 pandemic that resulted in 7 million deaths out of 770 million reported cases over the next four years. The global health emergency called for unprecedented efforts to monitor and reduce the rate of infection, pushing the study of new diagnostic methods. In this paper, we introduce a cheap, fast, and non-invasive detection system, which exploits only the exhaled breath. Specifically, provided an air sample, the mass spectra in the 10--351 mass-to-charge range are measured using an original nano-sampling device coupled with a high-precision spectrometer; then, the raw spectra are processed by custom software algorithms; the clean and augmented data are eventually classified using state-of-the-art machine-learning algorithms. An uncontrolled clinical trial was conducted between 2021 and 2022 on some 300 subjects who were concerned about being infected, either due to exhibiting symptoms or having quite recently recovered from illness. Despite the simplicity of use, our system showed a performance comparable to the traditional polymerase-chain-reaction and antigen testing in identifying cases of COVID-19 (that is, 0.95 accuracy, 0.94 recall, 0.96 specificity, and 0.92 F1-score). In light of these outcomes, we think that the proposed system holds the potential for substantial contributions to routine screenings and expedited responses during future epidemics, as it yields results comparable to state-of-the-art methods, providing them in a more rapid and less invasive manner.

cs.LG↗