arXiv subjects

Giuseppe Prencipe

Publications and source records attributed to Giuseppe Prencipe.

14 recordsLinked to original sources

arXiv2026

Safe Exploration of Arbitrary Dynamic Dangerous Networks

Given a team of agents on the nodes of a graph-based network, the exploration problem requires each node to be visited by at least one agent. In the classical distributed setting of static networks, agents do not know the topology of the network; in the more recently investigated setting of dynamic networks, they may have prior knowledge about the graph class (e.g., trees, tori, rings) or other parameters (e.g., number of nodes). Gotoh et al. (2021) are the first to study the exploration problem of arbitrary dynamic networks, under the necessary minimal assumption that any two nodes will be connected by a temporal path infinitely often (temporal connectivity assumption). In this paper, we extend their study of exploration under temporal connectivity by considering dangerous dynamic networks, i.e., containing possibly one or more black holes. Whenever an agent enters a black hole, it will be trapped forever. The safe exploration problem requires a team to explore all the safe nodes, ensuring that at least one agent will never be trapped in a black hole. We first prove that, given the necessary (and sufficient) number of agents, a team of oblivious agents can perpetually explore the safe nodes without any prior knowledge of the network or the team, without agent or node IDs, under semi-synchronous schedulers. Then, we provide an algorithm that enables agents to safely explore the network and terminate. In this case, agents are equipped with unique IDs and persistent memory, and they know the number of safe nodes. Yet, in both cases, we prove that it is impossible for a team of agents on dynamic networks to correctly mark only the ports leading to black holes.

cs.DC↗

arXiv2026

Adaptation Interfaces for In-Context Tabular Foundation Models in Time-to-Event Prediction

Tabular foundation models (TabFMs) achieve strong performance on structured data, particularly for standard classification and regression problems. Yet, extending them to censored time-to-event prediction is challenging because it requires properly handling censoring and event-time dynamics. Building on our prior work, we further link TabFMs with CoxPH and DeepHit and revise the context-resampled training procedure. We evaluate temporal zero-shot reformulation, classification-based fine-tuning, and survival-head adaptation using frozen TabFM backbones on 74 single-risk data sets, and we additionally study 4 competing-risk data sets. Zero-shot inference is effective on smaller single-risk data sets, whereas supervised adaptation becomes increasingly advantageous as data sets scale. Cox provides the most reliably strong interface, especially for Integrated Brier Score (IBS) on larger data sets. DeepHit is relatively stronger for the time-dependent Concordance Index than for IBS, while cause-specific MTLR ranks highest among the TabFM survival heads in the four-data-set competing-risk analysis. Classification fine-tuning becomes more competitive with zero-shot inference as data sets grow but remains weaker for probabilistic prediction. Overall, our results indicate that effective TabFM transfer depends on the data regime and on the statistical structure represented by the chosen adaptation interface. The implementation scripts used for this work are available at https://github.com/kaylode/survival-fm.

cs.LG↗

arXiv2026

One Human, $N$ Agents: Audit-Budget Allocation for LLM Agent Fleets under Miscalibrated, Correlated Confidence

A single human must audit $N$ LLM agents under a budget of $B \ll N$ audits per round, guided by self-reported confidence that may be adversarially miscalibrated and by correlated errors. We model this as budgeted noisy inspection over a two-level Gaussian copula and locate the miscalibration threshold $δ^*$ past which confidence-ranked auditing is \emph{worse} than random. Two a-priori expectations reverse: $δ^*$ \emph{rises} as the budget shrinks, and cross-family correlation is not low---shared difficulty dominates lineage. Five open-weight LLMs show operationally useless (near-constant) confidence, point estimates at or beyond the flip though CIs straddle it; a proprietary model is informative and lands below it. We give a quantitative criterion for \emph{vacuous} oversight, and replaying policies on recorded traces confirms the ordering.

cs.AI↗

arXiv2026

Black Hole Search by Scattered Agents in Dynamic Rings

In this paper, we address the challenge of locating a black hole within a dynamic graph using a set of scattered agents, which start from arbitrary positions in the graph. A black hole is defined as a node that silently eliminates any agent that visits it, effectively modeling network failures such as a crashed host or a destructive virus. The black hole search problem is considered solved when at least one agent survives and possesses a complete map of the graph, including the precise location of the black hole. Our study focuses on the scenario where the underlying graph is a dynamic 1-interval connected ring: a ring graph where, in each round, one edge may be absent. Agents communicate with other agents using movable pebbles that can be placed on nodes. In this setting, we demonstrate that three agents are sufficient to identify the black hole in $O(n^2)$ moves. Furthermore, we prove that this number of agents is optimal. Additionally, we establish that the complexity bound is tight, requiring $Ω(n^2)$ moves for any algorithm solving the problem with three agents, even when stronger communication mechanisms, such as unlimited-size whiteboards on nodes, are available.

cs.DC↗

arXiv2025

MIEO: encoding clinical data to enhance cardiovascular event prediction

As clinical data are becoming increasingly available, machine learning methods have been employed to extract knowledge from them and predict clinical events. While promising, approaches suffer from at least two main issues: low availability of labelled data and data heterogeneity leading to missing values. This work proposes the use of self-supervised auto-encoders to efficiently address these challenges. We apply our methodology to a clinical dataset from patients with ischaemic heart disease. Patient data is embedded in a latent space, built using unlabelled data, which is then used to train a neural network classifier to predict cardiovascular death. Results show improved balanced accuracy compared to applying the classifier directly to the raw data, demonstrating that this solution is promising, especially in conditions where availability of unlabelled data could increase.

cs.LG↗

arXiv2025

Universal Dancing by Luminous Robots under Sequential Schedulers

The Dancing problem requires a swarm of $n$ autonomous mobile robots to form a sequence of patterns, aka perform a choreography. Existing work has proven that some crucial restrictions on choreographies and initial configurations (e.g., on repetitions of patterns, periodicity, symmetries, contractions/expansions) must hold so that the Dancing problem can be solved under certain robot models. Here, we prove that these necessary constraints can be dropped by considering the LUMI model (i.e., where robots are endowed with a light whose color can be chosen from a constant-size palette) under the quite unexplored sequential scheduler. We formalize the class of Universal Dancing problems which require a swarm of $n$ robots starting from any initial configuration to perform a (periodic or finite) sequence of arbitrary patterns, only provided that each pattern consists of $n$ vertices (including multiplicities). However, we prove that, to be solvable under LUMI, the length of the feasible choreographies is bounded by the compositions of $n$ into the number of colors available to the robots. We provide an algorithm solving the Universal Dancing problem by exploiting the peculiar capability of sequential robots to implement a distributed counter mechanism. Even assuming non-rigid movements, our algorithm ensures spatial homogeneity of the performed choreography.

cs.DC↗

arXiv2021

Separating Bounded and Unbounded Asynchrony for Autonomous Robots: Point Convergence with Limited Visibility

Among fundamental problems in the context of distributed computing by autonomous mobile entities, one of the most representative and well studied is {\sc Point Convergence}: given an arbitrary initial configuration of identical entities, disposed in the Euclidean plane, move in such a way that, for all $\eps>0$, a configuration in which the separation between all entities is at most $\eps$ is eventually reached and maintained. The problem has been previously studied in a variety of settings, including full visibility, exact measurements (like distances and angles), and synchronous activation of entities. Our study concerns the minimal assumptions under which entities, moving asynchronously with limited and unknown visibility range and subject to limited imprecision in measurements, can be guaranteed to converge in this way. We present an algorithm that solves {\sc Point Convergence}, for entities in the plane, in such a setting, provided the degree of asynchrony is bounded: while any one entity is active, any other entity can be activated at most $k$ times, for some arbitrarily large but fixed $k$. This provides a strong positive answer to a decade old open question posed by Katreniak. We also prove that in a comparable setting that permits unbounded asynchrony, {\sc Point Convergence} in the plane is impossible, contingent on the natural assumption that algorithms maintain the (visible) connectivity among entities present in the initial configuration. This variant, that we call {\sc Cohesive Convergence}, serves to distinguish the power of bounded and unbounded asynchrony in the control of autonomous mobile entities, settling at the same time a long-standing question whether in the Euclidean plane synchronously scheduled entities are more powerful than asynchronously scheduled entities.

cs.DC↗

arXiv2020

Tight Bounds for Black Hole Search in Dynamic Rings

In this paper, we start the investigation of distributed computing by mobile agents in dangerous dynamic networks. The danger is posed by the presence in the network of a black hole BH, a harmful site that destroys all incoming agents without leaving any trace. The problem of determining the location of the black hole in a network, known as black hole search BHS, has been extensively studied in the literature, but always and only assuming that the network is static. At the same time, the existing results on mobile agents computing in dynamic networks never consider the presence of harmful sites. In this paper we start filling this research gap by studying black hole search in temporal rings, specifically focusing on 1-interval connectivity adversarial dynamics. The problem is solved if within finite time at least one agent survives and knows the location of BH. The main complexity parameters of BHS is the number of agents (called size) needed to solve the problem, and the number of moves (called cost) performed by the agents; in synchronous systems, such as temporal rings, an additional complexity measure is the amount of time until termination occurs. Feasibility and complexity depend on many parameters; in particular: whether the agents start from the same safe node or from possibly distinct safe locations, the size $n$ of the ring, whether or not $n$ is known, and the type of inter-agent communication (whiteboards, tokens, face-to-face, visual). In this paper, we provide a complete feasibility characterization for all instances of those parameters; all our algorithms are size optimal. Furthermore, we establish lower bounds on the cost (i.e., the number of moves) and time of size-optimal solutions for all instances of those parameters and show that our algorithms achieve those bound.

cs.DC↗

arXiv2017

Line-Recovery by Programmable Particles

Shape formation has been recently studied in distributed systems of programmable particles. In this paper we consider the shape recovery problem of restoring the shape when $f$ of the $n$ particles have crashed. We focus on the basic line shape, used as a tool for the construction of more complex configurations. We present a solution to the line recovery problem by the non-faulty anonymous particles; the solution works regardless of the initial distribution and number $f<n-4$ of faults, of the local orientations of the non-faulty entities, and of the number of non-faulty entities activated in each round (i.e., semi-synchronous adversarial scheduler).

cs.DC↗

arXiv2017

Gathering in Dynamic Rings

The gathering problem requires a set of mobile agents, arbitrarily positioned at different nodes of a network to group within finite time at the same location, not fixed in advanced. The extensive existing literature on this problem shares the same fundamental assumption: the topological structure does not change during the rendezvous or the gathering; this is true also for those investigations that consider faulty nodes. In other words, they only consider static graphs. In this paper we start the investigation of gathering in dynamic graphs, that is networks where the topology changes continuously and at unpredictable locations. We study the feasibility of gathering mobile agents, identical and without explicit communication capabilities, in a dynamic ring of anonymous nodes; the class of dynamics we consider is the classic 1-interval-connectivity. We focus on the impact that factors such as chirality (i.e., a common sense of orientation) and cross detection (i.e., the ability to detect, when traversing an edge, whether some agent is traversing it in the other direction), have on the solvability of the problem. We provide a complete characterization of the classes of initial configurations from which the gathering problem is solvable in presence and in absence of cross detection and of chirality. The feasibility results of the characterization are all constructive: we provide distributed algorithms that allow the agents to gather. In particular, the protocols for gathering with cross detection are time optimal. We also show that cross detection is a powerful computational element. We prove that, without chirality, knowledge of the ring size is strictly more powerful than knowledge of the number of agents; on the other hand, with chirality, knowledge of n can be substituted by knowledge of k, yielding the same classes of feasible initial configurations.

cs.DC↗

arXiv2017

Distributed Computing by Mobile Robots: Uniform Circle Formation

Consider a set of $n$ simple autonomous mobile robots (asynchronous, no common coordinate system, no identities, no central coordination, no direct communication, no memory of the past, non-rigid, deterministic) initially in distinct locations, moving freely in the plane and able to sense the positions of the other robots. We study the primitive task of the robots arranging themselves on the vertices of a regular $n$-gon not fixed in advance (Uniform Circle Formation). In the literature, the existing algorithmic contributions are limited to conveniently restricted sets of initial configurations of the robots and to more powerful robots. The question of whether such simple robots could deterministically form a uniform circle has remained open. In this paper, we constructively prove that indeed the Uniform Circle Formation problem is solvable for any initial configuration in which the robots are in distinct locations, without any additional assumption (if two robots are in the same location, the problem is easily seen to be unsolvable). In addition to closing a long-standing problem, the result of this paper also implies that, for pattern formation, asynchrony is not a computational handicap, and that additional powers such as chirality and rigidity are computationally irrelevant.

cs.DC↗

arXiv2015

Getting Close Without Touching: Near-Gathering for Autonomous Mobile Robots

In this paper we study the Near-Gathering problem for a finite set of dimensionless, deterministic, asynchronous, anonymous, oblivious and autonomous mobile robots with limited visibility moving in the Euclidean plane in Look-Compute-Move (LCM) cycles. In this problem, the robots have to get close enough to each other, so that every robot can see all the others, without touching (i.e., colliding with) any other robot. The importance of solving the Near-Gathering problem is that it makes it possible to overcome the restriction of having robots with limited visibility. Hence it allows to exploit all the studies (the majority, actually) done on this topic in the unlimited visibility setting. Indeed, after the robots get close enough to each other, they are able to see all the robots in the system, a scenario that is similar to the one where the robots have unlimited visibility. We present the first (deterministic) algorithm for the Near-Gathering problem, to the best of our knowledge, which allows a set of autonomous mobile robots to nearly gather within finite time without ever colliding. Our algorithm assumes some reasonable conditions on the input configuration (the Near-Gathering problem is easily seen to be unsolvable in general). Further, all the robots are assumed to have a compass (hence they agree on the "North" direction), but they do not necessarily have the same handedness (hence they may disagree on the clockwise direction). We also show how the robots can detect termination, i.e., detect when the Near-Gathering problem has been solved. This is crucial when the robots have to perform a generic task after having nearly gathered. We show that termination detection can be obtained even if the total number of robots is unknown to the robots themselves (i.e., it is not a parameter of the algorithm), and robots have no way to explicitly communicate.

cs.DC↗

arXiv2009

Phospholipid-Dextran with a Single Coupling Point: a Useful Amphiphile for Functionalization of Nanomaterials

Nanomaterials hold much promise for biological applications, but they require appropriate functionalization to provide biocompatibility in biological environments. For non-covalent functionalization with biocompatible polymers, the polymer must also remain attached to the nanomaterial after removal of its excess to mimic the high dilution conditions of administration in vivo. Reported here are the synthesis and utilization singly-substituted conjugates of dextran and a phospholipid (Dextran-DSPE) as stable coatings for nanomaterials. Suspensions of single walled carbon nanotubes were found not only to be stable to phosphate buffered saline (PBS), serum, and a variety of pHs after excess polymer removal, but also provide brighter photoluminescence than carbon nanotubes suspended by poly(ethylene glycol)-DSPE. In addition, both gold nanoparticles (AuNPs) and gold nanorods (AuNRs) were found to maintain their dispersion and characteristic optical absorbance after transfer into Dextran-DSPE, and were obtained in much better yield than similar suspensions with PEG-phospholipid and commonly used thiol-PEG. These suspensions were also stable to PBS, serum, and a variety of pHs after removal of excess polymer. Dextran-DSPE thus shows great promise as a general surfactant material for the functionalization of a variety of nanomaterials, which could facilitate future biological applications.

cond-mat.mtrl-sci↗

arXiv2009

PEG Branched Polymer for Functionalization of Nanomaterials with Ultralong Blood Circulation

Nanomaterials have been actively pursued for biological and medical applications in recent years. Here, we report the synthesis of several new poly(ethylene glycol) grafted branched-polymers for functionalization of various nanomaterials including carbon nanotubes, gold nanoparticles (NP) and gold nanorods (NRs), affording high aqueous solubility and stability for these materials. We synthesize different surfactant polymers based upon poly-(g-glutamic acid) (gPGA) and poly(maleic anhydride-alt-1-octadecene) (PMHC18). We use the abundant free carboxylic acid groups of gPGA for attaching lipophilic species such as pyrene or phospholipid, which bind to nanomaterials via robust physisorption. Additionally, the remaining carboxylic acids on gPGA or the amine-reactive anhydrides of PMHC18 are then PEGylated, providing extended hydrophilic groups, affording polymeric amphiphiles. We show that single-walled carbon nanotubes (SWNTs), Au NPs and NRs functionalized by the polymers exhibit high stability in aqueous solutions at different pHs, at elevated temperatures and in serum. Morever, the polymer-coated SWNTs exhibit remarkably long blood circulation (t1/2 22.1 h) upon intravenous injection into mice, far exceeding the previous record of 5.4 h. The ultra-long blood circulation time suggests greatly delayed clearance of nanomaterials by the reticuloendothelial system (RES) of mice, a highly desired property for in vivo applications of nanomaterials, including imaging and drug delivery.

cond-mat.mtrl-sci↗