Showing 1–11 of 11 results for author: Ferraz, L

Unified Uncertainty-Aware Diffusion for Multi-Agent Trajectory Modeling

Authors: Guillem Capellera, Antonio Rubio, Luis Ferraz, Antonio Agudo

Abstract: Multi-agent trajectory modeling has primarily focused on forecasting future states, often overlooking broader tasks like trajectory completion, which are crucial for real-world applications such as correcting tracking data. Existing methods also generally predict agents' states without offering any state-wise measure of uncertainty. Moreover, popular multi-modal sampling methods lack any error pro… ▽ More Multi-agent trajectory modeling has primarily focused on forecasting future states, often overlooking broader tasks like trajectory completion, which are crucial for real-world applications such as correcting tracking data. Existing methods also generally predict agents' states without offering any state-wise measure of uncertainty. Moreover, popular multi-modal sampling methods lack any error probability estimates for each generated scene under the same prior observations, making it difficult to rank the predictions during inference time. We introduce U2Diff, a \textbf{unified} diffusion model designed to handle trajectory completion while providing state-wise \textbf{uncertainty} estimates jointly. This uncertainty estimation is achieved by augmenting the simple denoising loss with the negative log-likelihood of the predicted noise and propagating latent space uncertainty to the real state space. Additionally, we incorporate a Rank Neural Network in post-processing to enable \textbf{error probability} estimation for each generated mode, demonstrating a strong correlation with the error relative to ground truth. Our method outperforms the state-of-the-art solutions in trajectory completion and forecasting across four challenging sports datasets (NBA, Basketball-U, Football-U, Soccer-U), highlighting the effectiveness of uncertainty and error probability estimation. Video at https://youtu.be/ngw4D4eJToE △ Less

Submitted 29 March, 2025; v1 submitted 24 March, 2025; originally announced March 2025.

Comments: Accepted to CVPR 2025 conference

Weak continuous measurements require more work than strong ones

Authors: Lorena Ballesteros Ferraz, Cyril Elouard

Abstract: We analyze a dynamical model for a quantum measurement process, able to capture nonideal (weak or inefficient) measurements. In this model, the irreversibility of the measurement dynamics is due to action of a reservoir at equilibrium, which cause decoherence between the states associated with different measurement results. We analyze the performance of measurement process generated by the model,… ▽ More We analyze a dynamical model for a quantum measurement process, able to capture nonideal (weak or inefficient) measurements. In this model, the irreversibility of the measurement dynamics is due to action of a reservoir at equilibrium, which cause decoherence between the states associated with different measurement results. We analyze the performance of measurement process generated by the model, by introducing figures of merits to quantify the strength of the measurement and its efficiency. We also derive and analyze a lower bound on the measurement work cost that we can relate to the measurement quality. We take as an illustration a qubit measurement owing to its coupling to a harmonic oscillator. We investigate the long sequences of extremely short and weak measurements (a.k.a continuous measurements), to find under which conditions they converge to an ndeal measurement and analyze their work cost. Surprisingly, we find that a sequence converging towards a projective measurement has a much larger work cost than a equivalent strong measurement obtained from a single intense interaction with the apparatus. We extend this result to a large class of models owing to scaling arguments. Our analysis offer new insights into the trade-offs between measurement strength, energy consumption, and information extraction in quantum measurement protocols. △ Less

Submitted 5 March, 2025; v1 submitted 13 February, 2025; originally announced February 2025.

TranSPORTmer: A Holistic Approach to Trajectory Understanding in Multi-Agent Sports

Authors: Guillem Capellera, Luis Ferraz, Antonio Rubio, Antonio Agudo, Francesc Moreno-Noguer

Abstract: Understanding trajectories in multi-agent scenarios requires addressing various tasks, including predicting future movements, imputing missing observations, inferring the status of unseen agents, and classifying different global states. Traditional data-driven approaches often handle these tasks separately with specialized models. We introduce TranSPORTmer, a unified transformer-based framework ca… ▽ More Understanding trajectories in multi-agent scenarios requires addressing various tasks, including predicting future movements, imputing missing observations, inferring the status of unseen agents, and classifying different global states. Traditional data-driven approaches often handle these tasks separately with specialized models. We introduce TranSPORTmer, a unified transformer-based framework capable of addressing all these tasks, showcasing its application to the intricate dynamics of multi-agent sports scenarios like soccer and basketball. Using Set Attention Blocks, TranSPORTmer effectively captures temporal dynamics and social interactions in an equivariant manner. The model's tasks are guided by an input mask that conceals missing or yet-to-be-predicted observations. Additionally, we introduce a CLS extra agent to classify states along soccer trajectories, including passes, possessions, uncontrolled states, and out-of-play intervals, contributing to an enhancement in modeling trajectories. Evaluations on soccer and basketball datasets show that TranSPORTmer outperforms state-of-the-art task-specific models in player forecasting, player forecasting-imputation, ball inference, and ball imputation. https://youtu.be/8VtSRm8oGoE △ Less

Submitted 9 November, 2024; v1 submitted 23 October, 2024; originally announced October 2024.

Comments: Accepted to ACCV 2024

What can PhD students and postdocs do to counter inequalities?

Authors: Lorena Ballesteros Ferraz, Carolina Charalambous, Sébastien Mouchet, Riccardo Muolo

Abstract: In this opinion article, we gathered some reflections and practical tips on what Early Stage Researchers do against inequalities in academia. This is the longer version of an opinion paper that was recently published on the website of the International Society for Optics and Photonics (spie.org), containing more details and suggestions. In this opinion article, we gathered some reflections and practical tips on what Early Stage Researchers do against inequalities in academia. This is the longer version of an opinion paper that was recently published on the website of the International Society for Optics and Photonics (spie.org), containing more details and suggestions. △ Less

Submitted 8 October, 2024; originally announced October 2024.

Comments: comments and feedback are welcome

Journal ref: https://spie.org/news/photonics-focus/septoct-2024/countering-inequalities-in-academia#_=_

FootBots: A Transformer-based Architecture for Motion Prediction in Soccer

Authors: Guillem Capellera, Luis Ferraz, Antonio Rubio, Antonio Agudo, Francesc Moreno-Noguer

Abstract: Motion prediction in soccer involves capturing complex dynamics from player and ball interactions. We present FootBots, an encoder-decoder transformer-based architecture addressing motion prediction and conditioned motion prediction through equivariance properties. FootBots captures temporal and social dynamics using set attention blocks and multi-attention block decoder. Our evaluation utilizes t… ▽ More Motion prediction in soccer involves capturing complex dynamics from player and ball interactions. We present FootBots, an encoder-decoder transformer-based architecture addressing motion prediction and conditioned motion prediction through equivariance properties. FootBots captures temporal and social dynamics using set attention blocks and multi-attention block decoder. Our evaluation utilizes two datasets: a real soccer dataset and a tailored synthetic one. Insights from the synthetic dataset highlight the effectiveness of FootBots' social attention mechanism and the significance of conditioned motion prediction. Empirical results on real soccer data demonstrate that FootBots outperforms baselines in motion prediction and excels in conditioned tasks, such as predicting the players based on the ball position, predicting the offensive (defensive) team based on the ball and the defensive (offensive) team, and predicting the ball position based on all players. Our evaluation connects quantitative and qualitative findings. https://youtu.be/9kaEkfzG3L8 △ Less

Submitted 28 June, 2024; originally announced June 2024.

Comments: Published as a conference paper at IEEE ICIP 2024

Leveraging modular values in quantum algorithms: the Deutsch-Jozsa

Authors: Lorena Ballesteros Ferraz, Timoteo Carletti, Yves Caudano

Abstract: We present a novel approach to quantum algorithms, by taking advantage of modular values, i.e., complex and unbounded quantities resulting from specific post-selected measurement scenarios. Our focus is on the problem of ascertaining whether a given function acting on a set of binary values is constant (uniformly yielding outputs of either all 0 or all 1), or balanced (a situation wherein half of… ▽ More We present a novel approach to quantum algorithms, by taking advantage of modular values, i.e., complex and unbounded quantities resulting from specific post-selected measurement scenarios. Our focus is on the problem of ascertaining whether a given function acting on a set of binary values is constant (uniformly yielding outputs of either all 0 or all 1), or balanced (a situation wherein half of the outputs are 0 and the other half are 1). Such problem can be solved by relying on the Deutsch-Jozsa algorithm. The proposed method, relying on the use of modular values, provides a high number of degrees of freedom for optimizing the new algorithm inspired from the Deutsch-Jozsa one. In particular, we explore meticulously the choices of the pre- and post-selected states. We eventually test the novel theoretical algorithm on a quantum computing platform. While the outcomes are currently not on par with the conventional approach, they nevertheless shed light on potential for future improvements, especially with less-optimized algorithms. We are thus confidend that the proposed proof of concept could prove its validity in bridging quantum algorithms and modular values research fields. △ Less

Submitted 10 June, 2024; originally announced June 2024.

On the relevance of weak measurements in dissipative quantum systems

Authors: Lorena Ballesteros Ferraz, John Martin, Yves Caudano

Abstract: We investigate the impact of dissipation on weak measurements. While weak measurements have been successful in signal amplification, dissipation can compromise their usefulness. More precisely, we show that in systems with non-degenerate eigenstates, weak values always converge to the expectation value of the measured observable as dissipation time tends to infinity, in contrast to systems with de… ▽ More We investigate the impact of dissipation on weak measurements. While weak measurements have been successful in signal amplification, dissipation can compromise their usefulness. More precisely, we show that in systems with non-degenerate eigenstates, weak values always converge to the expectation value of the measured observable as dissipation time tends to infinity, in contrast to systems with degenerate eigenstates, where the weak values can remain anomalous, i.e., outside the range of eigenvalues of the observable, even in the limit of an infinite dissipation time. In addition, we propose a method for extracting information about the dissipative dynamics of a system using weak values at short dissipation times. Specifically, we explore the amplification of the dissipation rate in a two-level system and the use of weak values to differentiate between Markovian and non-Markovian dissipative dynamics. We also find that weak measurements operating around a weak atom-cavity coupling can probe the atom dissipation through the weak value of non-Hermitian operators within the rotating-wave approximation of the weak interaction. △ Less

Submitted 1 August, 2023; originally announced August 2023.

Journal ref: Quantum Sci. Technol. 9 (2024) 035029

Revisiting weak values through non-normality

Authors: Lorena Ballesteros Ferraz, Riccardo Muolo, Yves Caudano, Timoteo Carletti

Abstract: Quantum measurement is one of the most fascinating and discussed phenomena in quantum physics, due to the impact on the system of the measurement action and the resulting interpretation issues. Scholars proposed weak measurements to amplify measured signals by exploiting a quantity called a weak value, but also to overcome philosophical difficulties related to the system perturbation induced by th… ▽ More Quantum measurement is one of the most fascinating and discussed phenomena in quantum physics, due to the impact on the system of the measurement action and the resulting interpretation issues. Scholars proposed weak measurements to amplify measured signals by exploiting a quantity called a weak value, but also to overcome philosophical difficulties related to the system perturbation induced by the measurement process. The method finds many applications and raises many philosophical questions as well, especially about the proper interpretation of the observations. In this paper, we show that any weak value can be expressed as the expectation value of a suitable non-normal operator. We propose a preliminary explanation of their anomalous and amplification behavior based on the theory of non-normal matrices and their link with non-normality: the weak value is different from an eigenvalue when the operator involved in the expectation value is non-normal. Our study paves the way for a deeper understanding of the measurement phenomenon, helps the design of experiments, and it is a call for collaboration to researchers in both fields to unravel new quantum phenomena induced by non-normality. △ Less

Submitted 2 November, 2023; v1 submitted 2 June, 2023; originally announced June 2023.

Journal ref: 2023 J. Phys. A: Math. Theor. 56 475303

Geometrical interpretation of the argument of Bargmann invariants and weak values in $N$-level quantum systems applying the Majorana symmetric representation

Authors: Lorena B Ferraz, Dominique L Lambert, Yves Caudano

Abstract: In this paper, we study the argument of weak values of general observables, succeeding to give a geometric description to this argument on the Bloch sphere. We apply the Majorana symmetric representation to reach this goal. The weak value of a general observable is proportional to the weak value of an effective projector: it arises from the normalized application of the observable over the initial… ▽ More In this paper, we study the argument of weak values of general observables, succeeding to give a geometric description to this argument on the Bloch sphere. We apply the Majorana symmetric representation to reach this goal. The weak value of a general observable is proportional to the weak value of an effective projector: it arises from the normalized application of the observable over the initial state, with a constant of proportionality that is real. The argument of the weak value of a projector on a pure state of an $N$-level system corresponds to a symplectic area in the complex projective space $(\text{CP}^{N-1})$, which can be represented geometrically with a sum of $N-1$ solid angles on the Bloch sphere using the Majorana stellar representation. Here, we show that the argument of the weak value of a general observable can be described, using Majorana representation, as the sum of $N-1$ solid angles on the Bloch sphere, merging both studies. These two approaches provide two geometrical descriptions, a first one in $\text{CP}^{N-1}$ and a second one on the Bloch sphere, after mapping the problem from the original space $(\text{CP}^{N-1})$ by making use of the Majorana representation. These results can also be applied to the argument of the third-order Bargmann invariant, the most fundamental order as the argument of any higher order invariant can be expressed as a sum of the argument of third-order Bargmann invariants. Finally, we focus on the argument of the weak value of a general spin-1 operator when its modulus diverges towards infinity. This divergence amplifies signals with great usefulness in experiments. △ Less

Submitted 10 November, 2022; originally announced November 2022.

Geometrical interpretation of the argument of weak values of general observables in N-level quantum systems

Authors: Lorena Ballesteros Ferraz, Dominique L. Lambert, Yves Caudano

Abstract: Observations in quantum weak measurements are determined by complex numbers called weak values. We present a geometrical interpretation of the argument of weak values of general Hermitian observables in $N$-dimensional quantum systems in terms of geometric phases. We formulate an arbitrary weak value in function of three real vectors on the unit sphere in $N^2-1$ dimensions, $S^{N^2-2}$. These vec… ▽ More Observations in quantum weak measurements are determined by complex numbers called weak values. We present a geometrical interpretation of the argument of weak values of general Hermitian observables in $N$-dimensional quantum systems in terms of geometric phases. We formulate an arbitrary weak value in function of three real vectors on the unit sphere in $N^2-1$ dimensions, $S^{N^2-2}$. These vectors are linked to the initial and final states, and to the weakly measured observable, respectively. We express pure states in the complex projective space of $N-1$ dimensions, $\mathbb{C}\textrm{P}^{N-1}$, which has a non-trivial representation as a $2N-2$ dimensional submanifold of $S^{N^2-2}$ (a generalization of the Bloch sphere for qudits). The argument of the weak value of a projector on a pure state of an $N$-level quantum system describes a geometric phase associated to the symplectic area of the geodesic triangle spanned by the vectors representing the pre-selected state, the projector and the post-selected state in $\mathbb{C}\textrm{P}^{N-1}$. We then proceed to show that the argument of the weak value of a general observable is equivalent to the argument of an effective Bargmann invariant. Hence, we extend the geometrical interpretation of projector weak values to weak values of general observables. In particular, we consider the generators of SU($N$) given by the generalized Gell-Mann matrices. Finally, we study in detail the case of the argument of weak values of general observables in two-level systems and we illustrate weak measurements in larger dimensional systems by considering projectors on degenerate subspaces, as well as Hermitian quantum gates. △ Less

Submitted 22 February, 2022; originally announced February 2022.

Comments: 29 pages, 3 figures

Journal ref: Quantum Sci. Technol. 7 (2022) 045028

Fracking Deep Convolutional Image Descriptors

Authors: Edgar Simo-Serra, Eduard Trulls, Luis Ferraz, Iasonas Kokkinos, Francesc Moreno-Noguer

Abstract: In this paper we propose a novel framework for learning local image descriptors in a discriminative manner. For this purpose we explore a siamese architecture of Deep Convolutional Neural Networks (CNN), with a Hinge embedding loss on the L2 distance between descriptors. Since a siamese architecture uses pairs rather than single image patches to train, there exist a large number of positive sample… ▽ More In this paper we propose a novel framework for learning local image descriptors in a discriminative manner. For this purpose we explore a siamese architecture of Deep Convolutional Neural Networks (CNN), with a Hinge embedding loss on the L2 distance between descriptors. Since a siamese architecture uses pairs rather than single image patches to train, there exist a large number of positive samples and an exponential number of negative samples. We propose to explore this space with a stochastic sampling of the training set, in combination with an aggressive mining strategy over both the positive and negative samples which we denote as "fracking". We perform a thorough evaluation of the architecture hyper-parameters, and demonstrate large performance gains compared to both standard CNN learning strategies, hand-crafted image descriptors like SIFT, and the state-of-the-art on learned descriptors: up to 2.5x vs SIFT and 1.5x vs the state-of-the-art in terms of the area under the curve (AUC) of the Precision-Recall curve. △ Less

Submitted 25 February, 2015; v1 submitted 19 December, 2014; originally announced December 2014.