arXiv subjects

Michele Flammini

Publications and source records attributed to Michele Flammini.

12 recordsLinked to original sources

arXiv2026

TrOCR for Medieval HTR: A Systematic Ablation Study with Cross-Dataset Validation

Fine-tuning transformer-based handwritten text recognition (HTR) models on medieval manuscripts is challenging because these models are pre-trained on modern text and must adapt to a very different visual domain. This paper studies how three controllable fine-tuning choices (contrast normalization, data augmentation, and layer freezing) affect recognition accuracy when adapting TrOCR to small historical datasets. We run controlled experiments on a 13th-century Italian manuscript (I-CT 91 "Cortonese") and replicate the same experimental grid on the public READ-16 benchmark as robustness evidence. On Cortonese, our best configuration achieves 8.03% character error rate (CER). Statistical comparisons across 13 configurations show that freezing up to three encoder layers or six decoder layers does not significantly harm accuracy, while deeper freezing becomes progressively detrimental. Removing contrast normalization (CLAHE) yields 7.84% CER, comparable to a domain-specialized baseline, suggesting strong optimization can reduce reliance on image preprocessing. Cross-dataset validation on READ-16 shows that decoder freezing thresholds transfer more robustly than encoder thresholds, and combined freezing strategies require dataset-specific re-validation. Finally, we use Grad-CAM gradient attributions and decoder cross-attention maps to diagnose error patterns and failure modes revealed by the ablations. Source code is available at https://github.com/LaudareProject/TrOCR-analysis

cs.CV↗

arXiv2025

Maximizing the Egalitarian Welfare in Friends and Enemies Games

We consider the complexity of maximizing egalitarian welfare in Friends and Enemies Games -- a subclass of hedonic games in which every agent partitions other agents into friends and enemies. We investigate two classic scenarios proposed in the literature, namely, Friends Appreciation ($\mathsf{FA}$) and Enemies Aversion ($\mathsf{EA}$): in the former, each agent primarily cares about the number of friends in her coalition, breaking ties based on the number of enemies, while in the latter, the opposite is true. For $\mathsf{EA}$, we show that our objective is hard to approximate within $O(n^{1-\epsilon})$, for any fixed $\epsilon>0$, and provide a polynomial-time $(n-1)$-approximation. For $\mathsf{FA}$, we obtain an NP-hardness result and a polynomial-time approximation algorithm. Our algorithm achieves a ratio of $2-\Theta(\frac{1}{n})$ when every agent has at least two friends; however, if some agent has at most one friend, its approximation ratio deteriorates to $n/2$. We recover the $2-\Theta(\frac{1}{n})$ approximation ratio for two important variants: when randomization is allowed and when the friendship relationship is symmetric. Additionally, for both $\mathsf{EA}$ and $\mathsf{FA}$ we identify special cases where the optimal egalitarian partition can be computed in polynomial time.

cs.GT↗

arXiv2025

Experimenting active and sequential learning in a medieval music manuscript

Optical Music Recognition (OMR) is a cornerstone of music digitization initiatives in cultural heritage, yet it remains limited by the scarcity of annotated data and the complexity of historical manuscripts. In this paper, we present a preliminary study of Active Learning (AL) and Sequential Learning (SL) tailored for object detection and layout recognition in an old medieval music manuscript. Leveraging YOLOv8, our system selects samples with the highest uncertainty (lowest prediction confidence) for iterative labeling and retraining. Our approach starts with a single annotated image and successfully boosts performance while minimizing manual labeling. Experimental results indicate that comparable accuracy to fully supervised training can be achieved with significantly fewer labeled examples. We test the methodology as a preliminary investigation on a novel dataset offered to the community by the Anonymous project, which studies laude, a poetical-musical genre spread across Italy during the 12th-16th Century. We show that in the manuscript at-hand, uncertainty-based AL is not effective and advocates for more usable methods in data-scarcity scenarios.

cs.CV↗

arXiv2025

Non-obvious Manipulability in Hedonic Games with Friends Appreciation Preferences

In this paper, we study non-obvious manipulability (NOM), a relaxed form of strategyproofness, in the context of Hedonic Games (HGs) with Friends Appreciation (FA) preferences. In HGs, the aim is to partition agents into coalitions according to their preferences which solely depend on the coalition they are assigned to. Under FA preferences, agents consider any other agent either a friend or an enemy, preferring coalitions with more friends and, in case of ties, the ones with fewer enemies. Our goal is to design mechanisms that prevent manipulations while optimizing social welfare. Prior research established that computing a welfare maximizing (optimum) partition for FA preferences is not strategyproof, and the best-known approximation to the optimum subject to strategyproofness is linear in the number of agents. In this work, we explore NOM to improve approximation results. We first prove the existence of a NOM mechanism that always outputs the optimum; however, we also demonstrate that the computation of an optimal partition is NP-hard. To address this complexity, we focus on approximation mechanisms and propose a NOM mechanism guaranteeing a $(4+o(1))$-approximation in polynomial time. Finally, we briefly discuss NOM in the case of Enemies Aversion (EA) preferences, the counterpart of FA, where agents give priority to coalitions with fewer enemies and show that no mechanism computing the optimum can be NOM.

cs.GT↗

arXiv2024

Fair Division with Social Impact

In this paper, we consider the problem of fair division of indivisible goods when the allocation of goods impacts society. Specifically, we introduce a second valuation function for each agent, determining the social impact of allocating a good to the agent. Such impact is considered desirable for the society -- the higher, the better. Our goal is to understand how to allocate goods fairly from the agents' perspective while maintaining society as happy as possible. To this end, we measure the impact on society using the utilitarian social welfare and provide both possibility and impossibility results. Our findings reveal that achieving good approximations, better than linear in the number of agents, is not possible while ensuring fairness to the agents. These impossibility results can be attributed to the fact that agents are completely unconscious of their social impact. Consequently, we explore scenarios where agents are socially aware, by introducing related fairness notions, and demonstrate that an appropriate definition of fairness aligns with the goal of maximizing the social objective.

cs.GT↗

arXiv2023

$\varepsilon$-fractional Core Stability in Hedonic Games

Hedonic Games (HGs) are a classical framework modeling coalition formation of strategic agents guided by their individual preferences. According to these preferences, it is desirable that a coalition structure (i.e. a partition of agents into coalitions) satisfies some form of stability. The most well-known and natural of such notions is arguably core-stability. Informally, a partition is core-stable if no subset of agents would like to deviate by regrouping in a so-called core-blocking coalition. Unfortunately, core-stable partitions seldom exist and even when they do, it is often computationally intractable to find one. To circumvent these problems, we propose the notion of $\varepsilon$-fractional core-stability, where at most an $\varepsilon$-fraction of all possible coalitions is allowed to core-block. It turns out that such a relaxation may guarantee both existence and polynomial-time computation. Specifically, we design efficient algorithms returning an $\varepsilon$-fractional core-stable partition, with $\varepsilon$ exponentially decreasing in the number of agents, for two fundamental classes of HGs: Simple Fractional and Anonymous. From a probabilistic point of view, being the definition of $\varepsilon$-fractional core equivalent to requiring that uniformly sampled coalitions core-block with probability lower than $\varepsilon$, we further extend the definition to handle more complex sampling distributions. Along this line, when valuations have to be learned from samples in a PAC-learning fashion, we give positive and negative results on which distributions allow the efficient computation of outcomes that are $\varepsilon$-fractional core-stable with arbitrarily high confidence.

cs.GT↗

arXiv2023

PAC learning and stabilizing Hedonic Games: towards a unifying approach

We study PAC learnability and PAC stabilizability of Hedonic Games (HGs), i.e., efficiently inferring preferences or core-stable partitions from samples. We first expand the known learnability/stabilizability landscape for some of the most prominent HGs classes, providing results for Friends and Enemies Games, Bottom Responsive, and Anonymous HGs. Then, having a broader view in mind, we attempt to shed light on the structural properties leading to learnability/stabilizability, or lack thereof, for specific HGs classes. Along this path, we focus on the fully expressive Hedonic Coalition Nets representation of HGs. We identify two sets of conditions that lead to efficient learnability, and which encompass all of the known positive learnability results. On the side of stability, we reveal that, while the freedom of choosing an ad hoc adversarial distribution is the most obvious hurdle to achieving PAC stability, it is not the only one. First, we show a distribution independent necessary condition for PAC stability. Then, we focus on $\W$-games, where players have individual preferences over other players and evaluate coalitions based on the least preferred member. We prove that these games are PAC stabilizable under the class of bounded distributions, which assign positive probability mass to all coalitions. Finally, we discuss why such a result is not easily extendable to other HGs classes even in this promising scenario. Namely, we establish a purely computational property necessary for achieving PAC stability.

cs.GT↗

arXiv2021

Unavailable Transit Feed Specification: Making it Available with Recurrent Neural Networks

Studies on public transportation in Europe suggest that European inhabitants use buses in ca. 56% of all public transport travels. One of the critical factors affecting such a percentage and more, in general, the demand for public transport services, with an increasing reluctance to use them, is their quality. End-users can perceive quality from various perspectives, including the availability of information, i.e., the access to details about the transit and the provided services. The approach proposed in this paper, using innovative methodologies resorting on data mining and machine learning techniques, aims to make available the unavailable data about public transport. In particular, by mining GPS traces, we manage to reconstruct the complete transit graph of public transport. The approach has been successfully validated on a real dataset collected from the local bus system of the city of L'Aquila (Italy). The experimental results demonstrate that the proposed approach and implemented framework are both effective and efficient, thus being ready for deployment.

cs.LG↗

arXiv2019

Parameterized Complexity of Manipulating Sequential Allocation

The sequential allocation protocol is a simple and popular mechanism to allocate indivisible goods, in which the agents take turns to pick the items according to a predefined sequence. While this protocol is not strategy-proof, it has been shown recently that finding a successful manipulation for an agent is an NP-hard problem (Aziz et al., 2017). Conversely, it is also known that finding an optimal manipulation can be solved in polynomial time in a few cases: if there are only two agents or if the manipulator has a binary or a lexicographic utility function. In this work, we take a parameterized approach to provide several new complexity results on this manipulation problem. More precisely, we give a complete picture of its parameterized complexity w.r.t. the following three parameters: the number $n$ of agents, the number $μ(a_1)$ of times the manipulator $a_1$ picks in the picking sequence, and the maximum range $\mathtt{rg}^{\max}$ of an item. This third parameter is a correlation measure on the preference rankings of the agents. In particular, we show that the problem of finding an optimal manipulation can be solved in polynomial time if $n$ or $μ(a_1)$ is a constant, and that it is fixed-parameter tractable w.r.t. $\mathtt{rg}^{\max}$ and $n+μ(a_1)$. Interestingly enough, we show that w.r.t. the single parameters $n$ and $μ(a_1)$ it is W[1]-hard. Moreover, we provide an integer program and a dynamic programming scheme to solve the manipulation problem and we show that a single manipulator can increase the utility of her bundle by a multiplicative factor which is at most 2.

cs.GT↗

arXiv2018

Almost Envy-Free Allocations with Connected Bundles

We study the existence of allocations of indivisible goods that are envy-free up to one good (EF1), under the additional constraint that each bundle needs to be connected in an underlying item graph. If the graph is a path and the utility functions are monotonic over bundles, we show the existence of EF1 allocations for at most four agents, and the existence of EF2 allocations for any number of agents; our proofs involve discrete analogues of the Stromquist's moving-knife protocol and the Su--Simmons argument based on Sperner's lemma. For identical utilities, we provide a polynomial-time algorithm that computes an EF1 allocation for any number of agents. For the case of two agents, we characterize the class of graphs that guarantee the existence of EF1 allocations as those whose biconnected components are arranged in a path; this property can be checked in linear time.

cs.GT↗

arXiv2017

Strategyproof Mechanisms for Additively Separable Hedonic Games and Fractional Hedonic Games

Additively separable hedonic games and fractional hedonic games have received considerable attention. They are coalition forming games of selfish agents based on their mutual preferences. Most of the work in the literature characterizes the existence and structure of stable outcomes (i.e., partitions in coalitions), assuming that preferences are given. However, there is little discussion on this assumption. In fact, agents receive different utilities if they belong to different partitions, and thus it is natural for them to declare their preferences strategically in order to maximize their benefit. In this paper we consider strategyproof mechanisms for additively separable hedonic games and fractional hedonic games, that is, partitioning methods without payments such that utility maximizing agents have no incentive to lie about their true preferences. We focus on social welfare maximization and provide several lower and upper bounds on the performance achievable by strategyproof mechanisms for general and specific additive functions. In most of the cases we provide tight or asymptotically tight results. All our mechanisms are simple and can be computed in polynomial time. Moreover, all the lower bounds are unconditional, that is, they do not rely on any computational or complexity assumptions.

cs.AI↗

arXiv2013

Approximating the Revenue Maximization Problem with Sharp Demands

We consider the revenue maximization problem with sharp multi-demand, in which $m$ indivisible items have to be sold to $n$ potential buyers. Each buyer $i$ is interested in getting exactly $d_i$ items, and each item $j$ gives a benefit $v_{ij}$ to buyer $i$. We distinguish between unrelated and related valuations. In the former case, the benefit $v_{ij}$ is completely arbitrary, while, in the latter, each item $j$ has a quality $q_j$, each buyer $i$ has a value $v_i$ and the benefit $v_{ij}$ is defined as the product $v_i q_j$. The problem asks to determine a price for each item and an allocation of bundles of items to buyers with the aim of maximizing the total revenue, that is, the sum of the prices of all the sold items. The allocation must be envy-free, that is, each buyer must be happy with her assigned bundle and cannot improve her utility. We first prove that, for related valuations, the problem cannot be approximated to a factor $O(m^{1-ε})$, for any $ε>0$, unless {\sf P} = {\sf NP} and that such result is asymptotically tight. In fact we provide a simple $m$-approximation algorithm even for unrelated valuations. We then focus on an interesting subclass of "proper" instances, that do not contain buyers a priori known not being able to receive any item. For such instances, we design an interesting $2$-approximation algorithm and show that no $(2-ε)$-approximation is possible for any $0<ε\leq 1$, unless {\sf P} $=$ {\sf NP}. We observe that it is possible to efficiently check if an instance is proper, and if discarding useless buyers is allowed, an instance can be made proper in polynomial time, without worsening the value of its optimal solution.

cs.GT↗