239700 research outputs found
Sort by
Implicit Neural Decision Trees
No full textRepresentation learning is a central topic in machine learning,
with significant efforts dedicated to encoding structured data such as
sequences, trees, and graphs for various downstream tasks. A branch of
these studies focuses on functional data analysis, which views data not
as discrete arrays but as continuous functions. When these functions
are parameterized using neural networks, they are called Implicit Neural
Representations (INR). INRs have been successfully applied to represent
diverse data types but, to the best of our knowledge, have not been used for
encoding decision models. This work addresses the novel challenge of using
INRs to represent decision trees. We introduce a tailored coordinate system
and train INRs to reconstruct decision trees with a loss function to minimize
node reconstruction errors. We benchmark implicit neural decision trees
on several datasets, showing that they can effectively represent individual
trees, and show potential extensions to tree forests through meta-learning
In medio stat virtus: lo spazio pubblico di connessione tra i poli universitari nella città di Pisa
No full text
A Narrative Review of Photon-Counting CT and Radiomics in Cardiothoracic Imaging: A Promising Match?
No full textInternational consensus statement on the use of ultrasound in hidradenitis suppurativa
No full textBackground: So far, the evidence on the use of ultrasound (US) in hidradenitis suppurativa (HS) demonstrates the utility of US in the diagnosis, scoring and assessment of HS; however, to date, there is no international consensus statement on the use of US in HS, and several published guidelines do not include this topic. Objectives: To create an international consensus statement on using US in HS that can cover and validate relevant aspects. Methods: A three-round Delphi study with a panel of international experts representing four continents and working with US in HS in their daily practice. The inclusion criteria of the experts and the set of questions in the survey were defined by a steering committee. A consensus of recommendation was defined when the percentage of agreement (sum of strongly agree and agree) was ε 70%. In between 50% and 69% of agreement, a suggestion was considered. Lower than 50% meant no consensus. Results: Twenty-four international experts from 14 countries participated in the study. A high percentage of consensus (96.4%) was achieved for important aspects of the use of US in HS, including the ultrasonographic indications, technical considerations, training, diagnostic criteria, staging systems, monitoring, support of US-guided procedures and planning of surgery, and the need for US in research and clinical trials. Conclusions: An international group of experts created a consensus statement with validated recommendations on the use of US in HS. Despite the challenges of the implementation of ultrasound in HS, this task force highly recommends the use of US in HS
Kibble-Zurek dynamics across the first-order quantum transitions of quantum Ising chains in the thermodynamic limit
No full textWe study the out-of-equilibrium Kibble-Zurek (KZ) dynamics in quantum Ising chains in a transverse field, driven by a time-dependent longitudinal field h(t ) = t/ts (ts is the timescale of the protocol), across their first-order quantum transitions (FOQTs) at h = 0. The KZ protocol starts at time ti < 0 from the negatively magnetized ground state for hi = ti/ts < 0. Then, the system evolves unitarily up to a time t f > 0, such that the magnetization of the state at time t f is positive. In finite-size systems, the KZ dynamics develops out-of-equilibrium finite-size scaling (OFSS) behaviors. Their scaling variables depend either exponentially or with a power law on the size, depending on the boundary conditions (BC). The OFSS functions can be computed in effective models restricted to appropriate low-energy (magnetized and/or kink) states. The KZ scaling behavior drastically changes in the thermodynamic limit (TL), defined as the infinite-size limit keeping t and ts fixed, which appears substantially unrelated with the OFSS regime, because it involves higher-energy multikink states, which are irrelevant in the OFSS limit. The numerical analyses of the KZ dynamics in the TL show the emergence of a quantum spinodal-like scaling behavior at the FOQTs for all considered BC, which is independent of the BC. The longitudinal magnetization changes sign at h(t ) = h_ > 0, where h_ decreases with increasing ts, as h* ∼ 1/ ln ts. Moreover, in the large-ts limit, the time dependence of the magnetization is described by a universal function of Ω = t/τs, with τs = ts/ ln ts
Melatonin improves the postharvest life of cut ruscus foliage after a long storage condition
No full textThis study aimed to investigate the effects of various pulse treatments on the postharvest longevity and physiological quality of cut Danae racemosa branches. The cut foliage was treated with deionised water (CTRL), 10 mu M thidiazuron (TDZ), 100 mu M melatonin (MEL), 300 mu M 5-aminolevulinic acid (AA), 300 mu M ascorbic acid (VIT C), and 300 mu M tocopherol (VIT E) for 24 h. The foliage was then subjected to simulated refrigerated transport (four weeks, at 6-8 degrees C in the dark) and retail conditions for up to 30 d (photoperiod: 12 h light/dark, light intensity: 50 mu mol m-2 s-1, temperature: 20-24 degrees C). Measurements included the water uptake rate, weight change, vase life, chlorophyll a fluorescence, colour index, chlorophyll content, ethylene production, and hormone balance. The results showed that TDZ or MEL treatments improved the water balance, delayed senescence, and enhanced photosynthetic efficiency, with higher chlorophyll concentration (1.3-1.6 g kg- 1 FW) and avoided ethylene increase (0.2-0.3 pL kg- 1 h- 1). Hormonal analysis revealed a decrease in ABA and an increase in cytokinin content in MEL-treated branches, indicating the role of MEL in modulating hormonal balance to extend vase life. These findings indicate that MEL is an effective postharvest treatment for preserving the visual and physiological quality of cut foliage, comparable to the well-established benefits of the TDZ
The Impact of Dance Movement Therapy on Subjective Well-Being and Quality of Life in Healthy Older Adults
No full textI diritti delle minoranze. Atti del Seminario interno al Corso di Alta Formazione in Giustizia costituzionale e tutela giurisdizionale dei diritti
No full textIl Volume raccoglie i contributi di alcuni dei partecipanti al Seminario
interno alla XIV edizione del Corso di Alta Formazione in
Giustizia costituzionale e tutela giurisdizionale dei diritti, organizzato
dal Dipartimento di Giurisprudenza dell’Università di Pisa nel
gennaio 2025.
Ancora una volta il CAF ha rappresentato un’occasione di confronto
proficuo sugli strumenti di tutela dei diritti fondamentali; in questo
contesto si collocano a pieno titolo le testimonianze portate dai
Corsisti su un tema messo a dura prova nel tempo attuale da tentativi,
spesso riusciti, di arretramento democratico, di recrudescenze
autoritarie e di avventurismi personalistici
Direct Localization With Direct-Path Interference for Reference Channel-Free Distributed Passive Radars
No full textThe unknown signals from non-cooperative illuminators of opportunity (IOs) and direct-path interference (DPI) make it challenging to achieve high-precision target localization for passive radars. Traditional dual-channel solutions heavily depend on high-quality reference channel (RC) observations, leading to increased hardware cost, computational overhead, and strict operational requirements. In this paper, a cost-effective signal model is proposed, which captures both noise and DPI-contaminated target echoes in an RC-free setting. Based on this model, a high-dimensional coupled parameter estimation problem is formulated, and a direct position estimator using an iterative optimization strategy with coordinate descent is developed to reduce computational complexity. The estimator iteratively subtracts the estimated DPI from the received signals and then updates the target position estimate. We also derive the Cramér–Rao lower bound (CRLB) to provide an accurate theoretical benchmark for the estimation accuracy. Numerical analysis is conducted to validate the effectiveness and robustness of the proposed estimator. The results demonstrate that the proposed estimator achieves a localization accuracy close to the CRLB and reveal that, although excessive DPI will result in performance degradation for target localization, moderate DPI can, in fact, allow to improve parameter estimation