Search

In this paper, we extend the investigation of biconservative surfaces with parallel normalized mean curvature vector fields (PNMC) in the 4-dimensional space forms, focusing on the hyperbolic space \mathbb{H}^4, the last remaining case to explore. We establish that an abstract surface admits a PNMC biconservative immersion in \mathbb{H}^4 if and only if it satisfies a certain intrinsic condition; if such an immersion exists, it is unique. We further analyze these abstract surfaces, showing that they form a two-parameter family. Additionally, we provide three characterizations of the intrinsic condition to explore the geometric properties of these surfaces., Comment: 23 pages. arXiv admin note: text overlap with arXiv:2211.08023

There is a growing interest in methods for detecting and interpreting changes in experimental time evolution data. Based on measured time series, the quantitative characterization of dynamical phase transitions at bifurcation points of the underlying chaotic systems is a notoriously difficult task. Building on prior theoretical studies that focus on the discontinuities at $q=1$ in the order-$q$ R\'enyi-entropy of the trajectory space, we measure the derivative of the spectrum. We derive within the general context of Markov processes a computationally efficient closed-form expression for this measure. We investigate its properties through well-known dynamical systems exploring its scope and limitations. The proposed mathematical instrument can serve as a predictor of dynamical phase transitions in time series., Comment: 11 pages, 3 figures

Smart meters, devices measuring the electricity and gas consumption of a household, are currently being deployed at a fast rate throughout the world. The data they collect are extremely useful, including in the fight against climate change. However, these data and the information that can be inferred from them are highly sensitive. Re-pseudonymization, i.e., the frequent replacement of random identifiers over time, is widely used to share smart meter data while mitigating the risk of re-identification. We here show how, in spite of re-pseudonymization, households' consumption records can be pieced together with high accuracy in large-scale datasets. We propose the first deep learning-based profiling attack against re-pseudonymized smart meter data. Our attack combines neural network embeddings, which are used to extract features from weekly consumption records and are tailored to the smart meter identification task, with a nearest neighbor classifier. We evaluate six neural networks architectures as the embedding model. Our results suggest that the Transformer and CNN-LSTM architectures vastly outperform previous methods as well as other architectures, successfully identifying the correct household 73.4% of the time among 5139 households based on electricity and gas consumption records (54.5% for electricity only). We further show that the features extracted by the embedding model maintain their effectiveness when transferred to a set of users disjoint from the one used to train the model. Finally, we extensively evaluate the robustness of our results. Taken together, our results strongly suggest that even frequent re-pseudonymization strategies can be reversed, strongly limiting their ability to prevent re-identification in practice., Comment: Extended version, including the Appendix, of a paper with the same title which will appear in the Proceedings of the Fourteenth ACM Conference on Data and Application Security and Privacy (CODASPY '24). The first two authors contributed equally

Highly integrated multichannel readout electronics is crucial in contemporary particle physics experiments. A novel silicon photomultiplier readout system based on the VMM3a ASIC was developed, for the first time exploiting this chip for calorimetric purposes. To extend the dynamic range the signal from each SiPM channel was processed by two electronics channels with different gain. A fully operational prototype system with 256 SiPM readout channels allowed the collection of data from a prototype of the ALICE Forward Hadron Calorimeter (FoCal-H). The design and the test beam results using high energy hadron beams are presented and discussed, confirming the applicability of VMM3a-based solutions for energy measurements in a high rate environment., Comment: 12 pages, 10 figures

Progress in machine learning (ML) has been fueled by scaling neural network models. This scaling has been enabled by ever more heroic feats of engineering, necessary for accommodating ML approaches that require high bandwidth communication between devices working in parallel. In this work, we propose a co-designed modular architecture and training approach for ML models, dubbed DIstributed PAth COmposition (DiPaCo). During training, DiPaCo distributes computation by paths through a set of shared modules. Together with a Local-SGD inspired optimization (DiLoCo) that keeps modules in sync with drastically reduced communication, Our approach facilitates training across poorly connected and heterogeneous workers, with a design that ensures robustness to worker failures and preemptions. At inference time, only a single path needs to be executed for each input, without the need for any model compression. We consider this approach as a first prototype towards a new paradigm of large-scale learning, one that is less synchronous and more modular. Our experiments on the widely used C4 benchmark show that, for the same amount of training steps but less wall-clock time, DiPaCo exceeds the performance of a 1 billion-parameter dense transformer language model by choosing one of 256 possible paths, each with a size of 150 million parameters.

In this work, we define the problem of finding an optimal query plan as finding spanning trees with low costs. This approach empowers the utilization of a series of spanning tree algorithms, thereby enabling systematic exploration of the plan search space over a join graph. Capitalizing on the polynomial time complexity of spanning tree algorithms, we present the Ensemble Spanning Tree Enumeration (ESTE) strategy. ESTE employs two conventional spanning tree algorithms, Prim's and Kruskal's, together to enhance the robustness of the query optimizer. In ESTE, multiple query plans are enumerated exploring different areas of the search space. This positions ESTE as an intermediate strategy between exhaustive and heuristic enumeration strategies. We show that ESTE is more robust in identifying efficient query plans for large queries. In the case of data modifications and workload demand increase, we believe that our approach can be a cheaper alternative to maintain optimizer robustness by integrating additional spanning tree algorithms rather than completely changing the optimizer to another plan enumeration algorithm. The experimental evaluation shows that ESTE achieves better consistency in plan quality and optimization time than existing solutions while identifying similarly optimal plans.

Sparsifying transforms became in the last decades widely known tools for finding structured sparse representations of signals in certain transform domains. Despite the popularity of classical transforms such as DCT and Wavelet, learning optimal transforms that guarantee good representations of data into the sparse domain has been recently analyzed in a series of papers. Typically, the conditioning number and representation ability are complementary key features of learning square transforms that may not be explicitly controlled in a given optimization model. Unlike the existing approaches from the literature, in our paper, we consider a new sparsifying transform model that enforces explicit control over the data representation quality and the condition number of the learned transforms. We confirm through numerical experiments that our model presents better numerical behavior than the state-of-the-art.

Local stochastic gradient descent (Local-SGD), also referred to as federated averaging, is an approach to distributed optimization where each device performs more than one SGD update per communication. This work presents an empirical study of {\it asynchronous} Local-SGD for training language models; that is, each worker updates the global parameters as soon as it has finished its SGD steps. We conduct a comprehensive investigation by examining how worker hardware heterogeneity, model size, number of workers, and optimizer could impact the learning performance. We find that with naive implementations, asynchronous Local-SGD takes more iterations to converge than its synchronous counterpart despite updating the (global) model parameters more frequently. We identify momentum acceleration on the global parameters when worker gradients are stale as a key challenge. We propose a novel method that utilizes a delayed Nesterov momentum update and adjusts the workers' local training steps based on their computation speed. This approach, evaluated with models up to 150M parameters on the C4 dataset, matches the performance of synchronous Local-SGD in terms of perplexity per update step, and significantly surpasses it in terms of wall clock time.

Economic and policy factors are driving the continuous increase in the adoption and usage of electrical vehicles (EVs). However, despite being a cleaner alternative to combustion engine vehicles, EVs have negative impacts on the lifespan of microgrid equipment and energy balance due to increased power demand and the timing of their usage. In our view grid management should leverage on EVs scheduling flexibility to support local network balancing through active participation in demand response programs. In this paper, we propose a model-free solution, leveraging Deep Q-Learning to schedule the charging and discharging activities of EVs within a microgrid to align with a target energy profile provided by the distribution system operator. We adapted the Bellman Equation to assess the value of a state based on specific rewards for EV scheduling actions and used a neural network to estimate Q-values for available actions and the epsilon-greedy algorithm to balance exploitation and exploration to meet the target energy profile. The results are promising showing that the proposed solution can effectively schedule the EVs charging and discharging actions to align with the target profile with a Person coefficient of 0.99, handling effective EVs scheduling situations that involve dynamicity given by the e-mobility features, relying only on data with no knowledge of EVs and microgrid dynamics., Comment: Submitted to journal

Interpreting traditional B-mode ultrasound images can be challenging due to image artifacts (e.g., shadowing, speckle), leading to low sensitivity and limited diagnostic accuracy. While Magnetic Resonance Imaging (MRI) has been proposed as a solution, it is expensive and not widely available. Furthermore, most biopsies are guided by Transrectal Ultrasound (TRUS) alone and can miss up to 52% cancers, highlighting the need for improved targeting. To address this issue, we propose ProsDectNet, a multi-task deep learning approach that localizes prostate cancer on B-mode ultrasound. Our model is pre-trained using radiologist-labeled data and fine-tuned using biopsy-confirmed labels. ProsDectNet includes a lesion detection and patch classification head, with uncertainty minimization using entropy to improve model performance and reduce false positive predictions. We trained and validated ProsDectNet using a cohort of 289 patients who underwent MRI-TRUS fusion targeted biopsy. We then tested our approach on a group of 41 patients and found that ProsDectNet outperformed the average expert clinician in detecting prostate cancer on B-mode ultrasound images, achieving a patient-level ROC-AUC of 82%, a sensitivity of 74%, and a specificity of 67%. Our results demonstrate that ProsDectNet has the potential to be used as a computer-aided diagnosis system to improve targeted biopsy and treatment planning., Comment: Accepted in NeurIPS 2023 (Medical Imaging meets NeurIPS Workshop)

Large language models (LLM) have become a critical component in many applications of machine learning. However, standard approaches to training LLM require a large number of tightly interconnected accelerators, with devices exchanging gradients and other intermediate states at each optimization step. While it is difficult to build and maintain a single computing cluster hosting many accelerators, it might be easier to find several computing clusters each hosting a smaller number of devices. In this work, we propose a distributed optimization algorithm, Distributed Low-Communication (DiLoCo), that enables training of language models on islands of devices that are poorly connected. The approach is a variant of federated averaging, where the number of inner steps is large, the inner optimizer is AdamW, and the outer optimizer is Nesterov momentum. On the widely used C4 dataset, we show that DiLoCo on 8 workers performs as well as fully synchronous optimization while communicating 500 times less. DiLoCo exhibits great robustness to the data distribution of each worker. It is also robust to resources becoming unavailable over time, and vice versa, it can seamlessly leverage resources that become available during training.

We present the performance of a full-length prototype of the ALICE Forward Calorimeter (FoCal). The detector is composed of a silicon-tungsten electromagnetic sampling calorimeter with longitudinal and transverse segmentation (FoCal-E) of about 20$X_0$ and a hadronic copper-scintillating-fiber calorimeter (FoCal-H) of about 5$\lambda_{\rm int}$. The data were taken between 2021 and 2023 at the CERN PS and SPS beam lines with hadron (electron) beams up to energies of 350 (300) GeV. Regarding FoCal-E, we report a comprehensive analysis of its response to minimum ionizing particles across all pad layers. The longitudinal shower profile of electromagnetic showers is measured with a layer-wise segmentation of 1$X_0$. As a projection to the performance of the final detector in electromagnetic showers, we demonstrate linearity in the full energy range, and show that the energy resolution fulfills the requirements for the physics needs. Additionally, the performance to separate two-showers events was studied by quantifying the transverse shower width. Regarding FoCal-H, we report a detailed analysis of the response to hadron beams between 60 and 350 GeV. The results are compared to simulations obtained with a Geant4 model of the test beam setup, which in particular for FoCal-E are in good agreement with the data. The energy resolution of FoCal-E was found to be lower than 3% at energies larger than 100 GeV. The response of FoCal-H to hadron beams was found to be linear, albeit with a significant intercept that is about factor 2 larger than in simulations. Its resolution, which is non-Gaussian and generally larger than in simulations, was quantified using the FWHM, and decreases from about 16% at 100 GeV to about 11% at 350 GeV. The discrepancy to simulations, which is particularly evident at low hadron energies, needs to be further investigated., Comment: 57 pages (without acronyms), 45 captioned figures

While loops are present in virtually all imperative programming languages. They are important both for practical reasons (performing a number of iterations not known in advance) and theoretical reasons (achieving Turing completeness). In this paper we propose an approach for incorporating while loops in an imperative language shallowly embedded in the Coq proof assistant. The main difficulty is that proving the termination of while loops is nontrivial, or impossible in the case of non-termination, whereas Coq only accepts programs endowed with termination proofs. Our solution is based on a new, general method for defining possibly non-terminating recursive functions in Coq. We illustrate the approach by proving termination and partial correctness of a program on linked lists., Comment: In Proceedings FROM 2023, arXiv:2309.12959

Tin perovskites are the most promising environmentally friendly alternative to lead perovskites. Among tin perovskites, FASnI3 (CH4N2SnI3) shows optimum band gap, and easy processability. However, the performance of FASnI3 based solar cells is incomparable to lead perovskites for several reasons, including energy band mismatch between the perovskite absorber film and the charge transporting layers (CTLs) for both types of carriers, i.e., for electrons (ETLs) and holes (HTLs). However, the band diagrams in the literature are inconsistent, and the charge extraction dynamics are poorly understood. In this paper, we study the energy band positions of FASnI3 based perovskites using Kelvin probe (KP) and photoelectron yield spectroscopy (PYS) to provide a precise band diagram of the most used device stack. In addition, we analyze the defects within the current energetic landscape of tin perovskites. We uncover the role of bathocuproine (BCP) in enhancing the electron extraction at the fullerene C60/BCP interface. Furthermore, we used transient surface photovoltage (tr-SPV) for the first time for tin perovskites to understand the charge extraction dynamics of the most reported HTLs such as NiOx and PEDOT, and ETLs such as C60, ICBA, and PCBM. Finally, we used Hall effect, KP, and time-resolved photoluminescence (TRPL) to estimate an accurate value of the p-doping concentration in FASnI3 and showed a consistent result of 1.5 * 1017 cm-3. Our findings prove that the energetic system of tin halide perovskites is deformed and should be redesigned independently from lead perovskites to unlock the full potential of tin perovskites., Comment: 22 pages, 5 figures

Solution-processable near-infrared (NIR) photodetectors are urgently needed for a wide range of next-generation electronics, including sensors, optical communications and bioimaging. However, there is currently a compromise between low toxicity and slow (<300 kHz cut-off frequency) organic materials versus faster detectors (>300 kHz cut-off frequency) based on compounds containing toxic lead or cadmium. Herein, we circumvent this trade-off by developing solution-processed AgBiS2 photodetectors with high cut-off frequencies under both white light (>1 MHz) and NIR (approaching 500 kHz) illumination. These high cut-off frequencies are due to the short transit distances of charge-carriers in the AgBiS2 photodetectors, which arise from the strong light absorption of these materials, such that film thicknesses well below 120 nm are adequate to absorb >65% of near-infrared to visible light. By finely controlling the thickness of the photoactive layer, we can modulate the charge-collection efficiency, achieve low dark current densities, and minimize the effects of ion migration to realize fast photodetectors that are stable in air. These outstanding characteristics enable real-time heartbeat sensors based on NIR AgBiS2 photodetectors. # equal contribution, * corresponding authors, Comment: 29 pages, 4 figures, 1 table

This paper presents a kernelized version of the t-SNE algorithm, capable of mapping high-dimensional data to a low-dimensional space while preserving the pairwise distances between the data points in a non-Euclidean metric. This can be achieved using a kernel trick only in the high dimensional space or in both spaces, leading to an end-to-end kernelized version. The proposed kernelized version of the t-SNE algorithm can offer new views on the relationships between data points, which can improve performance and accuracy in particular applications, such as classification problems involving kernel methods. The differences between t-SNE and its kernelized version are illustrated for several datasets, showing a neater clustering of points belonging to different classes.