Showing 1–50 of 50 results for author: Corso, G

DisCo-Diff: Enhancing Continuous Diffusion Models with Discrete Latents

Authors: Yilun Xu, Gabriele Corso, Tommi Jaakkola, Arash Vahdat, Karsten Kreis

Abstract: Diffusion models (DMs) have revolutionized generative learning. They utilize a diffusion process to encode data into a simple Gaussian distribution. However, encoding a complex, potentially multimodal data distribution into a single continuous Gaussian distribution arguably represents an unnecessarily challenging learning problem. We propose Discrete-Continuous Latent Variable Diffusion Models (Di… ▽ More Diffusion models (DMs) have revolutionized generative learning. They utilize a diffusion process to encode data into a simple Gaussian distribution. However, encoding a complex, potentially multimodal data distribution into a single continuous Gaussian distribution arguably represents an unnecessarily challenging learning problem. We propose Discrete-Continuous Latent Variable Diffusion Models (DisCo-Diff) to simplify this task by introducing complementary discrete latent variables. We augment DMs with learnable discrete latents, inferred with an encoder, and train DM and encoder end-to-end. DisCo-Diff does not rely on pre-trained networks, making the framework universally applicable. The discrete latents significantly simplify learning the DM's complex noise-to-data map** by reducing the curvature of the DM's generative ODE. An additional autoregressive transformer models the distribution of the discrete latents, a simple step because DisCo-Diff requires only few discrete variables with small codebooks. We validate DisCo-Diff on toy data, several image synthesis tasks as well as molecular docking, and find that introducing discrete latents consistently improves model performance. For example, DisCo-Diff achieves state-of-the-art FID scores on class-conditioned ImageNet-64/128 datasets with ODE sampler. △ Less

Submitted 3 July, 2024; originally announced July 2024.

Comments: project page: https://research.nvidia.com/labs/lpr/disco-diff

Can Ephapticity Contributes to the Brain Complexity?

Authors: Gabriel Moreno Cunha, Gillberto Corso, Matheus Phellipe Brasil de Sousa, Gustavo Zampier dos Santos Lima

Abstract: The inquiry into the origin of brain complexity remains a pivotal question in neuroscience. While synaptic stimuli are acknowledged as significant, their efficacy often falls short in elucidating the extensive interconnections of the brain and nuanced levels of cognitive integration. Recent advances in neuroscience have brought the mechanisms underlying the generation of highly intricate dynamics,… ▽ More The inquiry into the origin of brain complexity remains a pivotal question in neuroscience. While synaptic stimuli are acknowledged as significant, their efficacy often falls short in elucidating the extensive interconnections of the brain and nuanced levels of cognitive integration. Recent advances in neuroscience have brought the mechanisms underlying the generation of highly intricate dynamics, emergent patterns, and sophisticated oscillatory signals into question. Within this context, our study, in alignment with current research, posits the hypothesis that ephaptic communication may emerge as the primary candidate for unraveling optimal brain complexity. In this investigation, we conducted a comparative analysis between two types of networks utilizing the Quadratic Integrate-and-Fire Ephaptic model (QIF-E): (I) a small-world synaptic network (ephaptic-off) and (II) a mixed composite network comprising a small-world synaptic network with the addition of an ephaptic network (ephaptic-on). Utilizing the Multiscale Entropy methodology, we conducted an in-depth analysis of the responses generated by both network configurations, with complexity assessed by integrating across all temporal scales. Our findings demonstrate that ephaptic coupling enhances complexity under specific topological conditions, considering variables such as time, spatial scales, and synaptic intensity. These results offer fresh insights into the dynamics of communication within the nervous system and underscore the fundamental role of ephapticity in regulating complex brain functions. △ Less

Submitted 25 April, 2024; originally announced April 2024.

Quantum Subroutine for Variance Estimation: Algorithmic Design and Applications

Authors: Anna Bernasconi, Alessandro Berti, Gianna M. Del Corso, Riccardo Guidotti, Alessandro Poggiali

Abstract: Quantum computing sets the foundation for new ways of designing algorithms, thanks to the peculiar properties inherited by quantum mechanics. The exploration of this new paradigm faces new challenges concerning which field quantum speedup can be achieved. Towards finding solutions, looking for the design of quantum subroutines that are more efficient than their classical counterpart poses solid pi… ▽ More Quantum computing sets the foundation for new ways of designing algorithms, thanks to the peculiar properties inherited by quantum mechanics. The exploration of this new paradigm faces new challenges concerning which field quantum speedup can be achieved. Towards finding solutions, looking for the design of quantum subroutines that are more efficient than their classical counterpart poses solid pillars to new powerful quantum algorithms. Herewith, we delve into a grounding subroutine, the computation of the variance, whose usefulness spaces across different fields of application, particularly the Artificial Intelligence (AI) one. Indeed, the finding of the quantum counterpart of these building blocks impacts vertically those algorithms that leverage this metric. In this work, we propose QVAR, a quantum subroutine, to compute the variance that exhibits a logarithmic complexity both in the circuit depth and width, excluding the state preparation cost. With the vision of showing the use of QVAR as a subroutine for new quantum algorithms, we tackle two tasks from the AI domain: Feature Selection and Outlier Detection. In particular, we showcase two AI hybrid quantum algorithms that leverage QVAR: the Hybrid Quantum Feature Selection (HQFS) algorithm and the Quantum Outlier Detection Algorithm (QODA). In this manuscript, we describe the implementation of QVAR, HQFS, and QODA, providing their correctness and complexities and showing the effectiveness of these hybrid quantum algorithms with respect to their classical counterpart. △ Less

Submitted 26 February, 2024; originally announced March 2024.

Deep Confident Steps to New Pockets: Strategies for Docking Generalization

Authors: Gabriele Corso, Arthur Deng, Benjamin Fry, Nicholas Polizzi, Regina Barzilay, Tommi Jaakkola

Abstract: Accurate blind docking has the potential to lead to new biological breakthroughs, but for this promise to be realized, docking methods must generalize well across the proteome. Existing benchmarks, however, fail to rigorously assess generalizability. Therefore, we develop DockGen, a new benchmark based on the ligand-binding domains of proteins, and we show that existing machine learning-based dock… ▽ More Accurate blind docking has the potential to lead to new biological breakthroughs, but for this promise to be realized, docking methods must generalize well across the proteome. Existing benchmarks, however, fail to rigorously assess generalizability. Therefore, we develop DockGen, a new benchmark based on the ligand-binding domains of proteins, and we show that existing machine learning-based docking models have very weak generalization abilities. We carefully analyze the scaling laws of ML-based docking and show that, by scaling data and model size, as well as integrating synthetic data strategies, we are able to significantly increase the generalization capacity and set new state-of-the-art performance across benchmarks. Further, we propose Confidence Bootstrap**, a new training paradigm that solely relies on the interaction between diffusion and confidence models and exploits the multi-resolution generation process of diffusion models. We demonstrate that Confidence Bootstrap** significantly improves the ability of ML-based docking methods to dock to unseen protein classes, edging closer to accurate and generalizable blind docking methods. △ Less

Submitted 28 February, 2024; originally announced February 2024.

Journal ref: International Conference on Learning Representations 2024

Dirichlet Flow Matching with Applications to DNA Sequence Design

Authors: Hannes Stark, Bowen **g, Chenyu Wang, Gabriele Corso, Bonnie Berger, Regina Barzilay, Tommi Jaakkola

Abstract: Discrete diffusion or flow models could enable faster and more controllable sequence generation than autoregressive models. We show that naïve linear flow matching on the simplex is insufficient toward this goal since it suffers from discontinuities in the training target and further pathologies. To overcome this, we develop Dirichlet flow matching on the simplex based on mixtures of Dirichlet dis… ▽ More Discrete diffusion or flow models could enable faster and more controllable sequence generation than autoregressive models. We show that naïve linear flow matching on the simplex is insufficient toward this goal since it suffers from discontinuities in the training target and further pathologies. To overcome this, we develop Dirichlet flow matching on the simplex based on mixtures of Dirichlet distributions as probability paths. In this framework, we derive a connection between the mixtures' scores and the flow's vector field that allows for classifier and classifier-free guidance. Further, we provide distilled Dirichlet flow matching, which enables one-step sequence generation with minimal performance hits, resulting in $O(L)$ speedups compared to autoregressive models. On complex DNA sequence generation tasks, we demonstrate superior performance compared to all baselines in distributional metrics and in achieving desired design targets for generated sequences. Finally, we show that our classifier-free guidance approach improves unconditional generation and is effective for generating DNA that satisfies design targets. Code is available at https://github.com/HannesStark/dirichlet-flow-matching. △ Less

Submitted 30 May, 2024; v1 submitted 8 February, 2024; originally announced February 2024.

Comments: Published at ICML 2024. (Proceedings of the 41st International Conference on Machine Learning, Vienna, Austria. PMLR 235, 2024)

Bayesian Time-Lapse Full Waveform Inversion using Hamiltonian Monte Carlo

Authors: Paulo Douglas S. de Lima, Mauro S. Ferreira, Gilberto Corso, João M. de Araújo

Abstract: Time-lapse images carry out important information about dynamic changes in Earth's interior which can be inferred using different Full Waveform Inversion (FWI) schemes. The estimation process is performed by manipulating more than one seismic dataset, associated with the baseline and monitors surveys. The time-lapse variations can be so minute and localised that quantifying the uncertainties becom… ▽ More Time-lapse images carry out important information about dynamic changes in Earth's interior which can be inferred using different Full Waveform Inversion (FWI) schemes. The estimation process is performed by manipulating more than one seismic dataset, associated with the baseline and monitors surveys. The time-lapse variations can be so minute and localised that quantifying the uncertainties becomes fundamental to assessing the reliability of the results. The Bayesian formulation of the FWI problem naturally provides confidence levels in the solution, but evaluating the uncertainty of time-lapse seismic inversion remains a challenge due to the ill-posedness and high dimensionality of the problem. The Hamiltonian Monte Carlo (HMC) can be used to effectively sample over high dimensional distributions with affordable computational efforts. In this context, we propose a probabilistic Bayesian sequential approach for time-lapse FWI using the HMC method. Our approach relies on the integration of the baseline survey information as prior knowledge in the monitor estimation. We compare the proposed methodology with a parallel scheme in perfect and perturbed acquisition geometry scenarios. We also investigate the correlation effect between baseline and monitor samples in the propagated uncertainties. The results show that our strategy provides accurate times-lapse estimates with errors of similar magnitude to the parallel methodology. △ Less

Submitted 9 December, 2023; v1 submitted 6 November, 2023; originally announced November 2023.

Comments: 12 pages, 6 figures

Particle Guidance: non-I.I.D. Diverse Sampling with Diffusion Models

Authors: Gabriele Corso, Yilun Xu, Valentin de Bortoli, Regina Barzilay, Tommi Jaakkola

Abstract: In light of the widespread success of generative models, a significant amount of research has gone into speeding up their sampling time. However, generative models are often sampled multiple times to obtain a diverse set incurring a cost that is orthogonal to sampling time. We tackle the question of how to improve diversity and sample efficiency by moving beyond the common assumption of independen… ▽ More In light of the widespread success of generative models, a significant amount of research has gone into speeding up their sampling time. However, generative models are often sampled multiple times to obtain a diverse set incurring a cost that is orthogonal to sampling time. We tackle the question of how to improve diversity and sample efficiency by moving beyond the common assumption of independent samples. We propose particle guidance, an extension of diffusion-based generative sampling where a joint-particle time-evolving potential enforces diversity. We analyze theoretically the joint distribution that particle guidance generates, how to learn a potential that achieves optimal diversity, and the connections with methods in other disciplines. Empirically, we test the framework both in the setting of conditional image generation, where we are able to increase diversity without affecting quality, and molecular conformer generation, where we reduce the state-of-the-art median error by 13% on average. △ Less

Submitted 24 November, 2023; v1 submitted 19 October, 2023; originally announced October 2023.

DiffDock-PP: Rigid Protein-Protein Docking with Diffusion Models

Authors: Mohamed Amine Ketata, Cedrik Laue, Ruslan Mammadov, Hannes Stärk, Menghua Wu, Gabriele Corso, Céline Marquet, Regina Barzilay, Tommi S. Jaakkola

Abstract: Understanding how proteins structurally interact is crucial to modern biology, with applications in drug discovery and protein design. Recent machine learning methods have formulated protein-small molecule docking as a generative problem with significant performance boosts over both traditional and deep learning baselines. In this work, we propose a similar approach for rigid protein-protein docki… ▽ More Understanding how proteins structurally interact is crucial to modern biology, with applications in drug discovery and protein design. Recent machine learning methods have formulated protein-small molecule docking as a generative problem with significant performance boosts over both traditional and deep learning baselines. In this work, we propose a similar approach for rigid protein-protein docking: DiffDock-PP is a diffusion generative model that learns to translate and rotate unbound protein structures into their bound conformations. We achieve state-of-the-art performance on DIPS with a median C-RMSD of 4.85, outperforming all considered baselines. Additionally, DiffDock-PP is faster than all search-based methods and generates reliable confidence estimates for its predictions. Our code is publicly available at $\texttt{https://github.com/ketatam/DiffDock-PP}$ △ Less

Submitted 7 April, 2023; originally announced April 2023.

Comments: ICLR Machine Learning for Drug Discovery (MLDD) Workshop 2023

EigenFold: Generative Protein Structure Prediction with Diffusion Models

Authors: Bowen **g, Ezra Erives, Peter Pao-Huang, Gabriele Corso, Bonnie Berger, Tommi Jaakkola

Abstract: Protein structure prediction has reached revolutionary levels of accuracy on single structures, yet distributional modeling paradigms are needed to capture the conformational ensembles and flexibility that underlie biological function. Towards this goal, we develop EigenFold, a diffusion generative modeling framework for sampling a distribution of structures from a given protein sequence. We defin… ▽ More Protein structure prediction has reached revolutionary levels of accuracy on single structures, yet distributional modeling paradigms are needed to capture the conformational ensembles and flexibility that underlie biological function. Towards this goal, we develop EigenFold, a diffusion generative modeling framework for sampling a distribution of structures from a given protein sequence. We define a diffusion process that models the structure as a system of harmonic oscillators and which naturally induces a cascading-resolution generative process along the eigenmodes of the system. On recent CAMEO targets, EigenFold achieves a median TMScore of 0.84, while providing a more comprehensive picture of model uncertainty via the ensemble of sampled structures relative to existing methods. We then assess EigenFold's ability to model and predict conformational heterogeneity for fold-switching proteins and ligand-induced conformational change. Code is available at https://github.com/b**g2016/EigenFold. △ Less

Submitted 4 April, 2023; originally announced April 2023.

Comments: ICLR MLDD workshop 2023

Modeling Molecular Structures with Intrinsic Diffusion Models

Authors: Gabriele Corso

Abstract: Since its foundations, more than one hundred years ago, the field of structural biology has strived to understand and analyze the properties of molecules and their interactions by studying the structure that they take in 3D space. However, a fundamental challenge with this approach has been the dynamic nature of these particles, which forces us to model not a single but a whole distribution of str… ▽ More Since its foundations, more than one hundred years ago, the field of structural biology has strived to understand and analyze the properties of molecules and their interactions by studying the structure that they take in 3D space. However, a fundamental challenge with this approach has been the dynamic nature of these particles, which forces us to model not a single but a whole distribution of structures for every molecular system. This thesis proposes Intrinsic Diffusion Modeling, a novel approach to this problem based on combining diffusion generative models with scientific knowledge about the flexibility of biological complexes. The knowledge of these degrees of freedom is translated into the definition of a manifold over which the diffusion process is defined. This manifold significantly reduces the dimensionality and increases the smoothness of the generation space allowing for significantly faster and more accurate generative processes. We demonstrate the effectiveness of this approach on two fundamental tasks at the basis of computational chemistry and biology: molecular conformer generation and molecular docking. In both tasks, we construct the first deep learning method to outperform traditional computational approaches achieving an unprecedented level of accuracy for scalable programs. △ Less

Submitted 22 February, 2023; originally announced February 2023.

Comments: MIT Master of Science thesis. arXiv admin note: substantial text overlap with arXiv:2210.01776, arXiv:2206.01729

Quantum Hitting Time according to a given distribution

Authors: P. Boito, G. M. Del Corso

Abstract: In this work we focus on the notion of quantum hitting time for discrete-time Szegedy quantum walks, compared to its classical counterpart. Under suitable hypotheses, quantum hitting time is known to be of the order of the square root of classical hitting time: this quadratic speedup is a remarkable example of the computational advantages associated with quantum approaches. Our purpose here is t… ▽ More In this work we focus on the notion of quantum hitting time for discrete-time Szegedy quantum walks, compared to its classical counterpart. Under suitable hypotheses, quantum hitting time is known to be of the order of the square root of classical hitting time: this quadratic speedup is a remarkable example of the computational advantages associated with quantum approaches. Our purpose here is twofold. On one hand, we provide a detailed proof of quadratic speedup for time-reversible walks within the Szegedy framework, in a language that should be familiar to the linear algebra community. Moreover, we explore the use of a general distribution in place of the stationary distribution in the definition of quantum hitting time, through theoretical considerations and numerical experiments. △ Less

Submitted 17 February, 2023; originally announced February 2023.

MSC Class: 81P45; 65F99

Quantum Clustering with k-Means: a Hybrid Approach

Authors: Alessandro Poggiali, Alessandro Berti, Anna Bernasconi, Gianna M. Del Corso, Riccardo Guidotti

Abstract: Quantum computing is a promising paradigm based on quantum theory for performing fast computations. Quantum algorithms are expected to surpass their classical counterparts in terms of computational complexity for certain tasks, including machine learning. In this paper, we design, implement, and evaluate three hybrid quantum k-Means algorithms, exploiting different degree of parallelism. Indeed, e… ▽ More Quantum computing is a promising paradigm based on quantum theory for performing fast computations. Quantum algorithms are expected to surpass their classical counterparts in terms of computational complexity for certain tasks, including machine learning. In this paper, we design, implement, and evaluate three hybrid quantum k-Means algorithms, exploiting different degree of parallelism. Indeed, each algorithm incrementally leverages quantum parallelism to reduce the complexity of the cluster assignment step up to a constant cost. In particular, we exploit quantum phenomena to speed up the computation of distances. The core idea is that the computation of distances between records and centroids can be executed simultaneously, thus saving time, especially for big datasets. We show that our hybrid quantum k-Means algorithms can be more efficient than the classical version, still obtaining comparable clustering results. △ Less

Submitted 15 December, 2022; v1 submitted 13 December, 2022; originally announced December 2022.

Report number: 2212.06691

Journal ref: Theoretical Computer Science 2024

Learning Graph Search Heuristics

Authors: Michal Pándy, Weikang Qiu, Gabriele Corso, Petar Veličković, Rex Ying, Jure Leskovec, Pietro Liò

Abstract: Searching for a path between two nodes in a graph is one of the most well-studied and fundamental problems in computer science. In numerous domains such as robotics, AI, or biology, practitioners develop search heuristics to accelerate their pathfinding algorithms. However, it is a laborious and complex process to hand-design heuristics based on the problem and the structure of a given use case. H… ▽ More Searching for a path between two nodes in a graph is one of the most well-studied and fundamental problems in computer science. In numerous domains such as robotics, AI, or biology, practitioners develop search heuristics to accelerate their pathfinding algorithms. However, it is a laborious and complex process to hand-design heuristics based on the problem and the structure of a given use case. Here we present PHIL (Path Heuristic with Imitation Learning), a novel neural architecture and a training algorithm for discovering graph search and navigation heuristics from data by leveraging recent advances in imitation learning and graph representation learning. At training time, we aggregate datasets of search trajectories and ground-truth shortest path distances, which we use to train a specialized graph neural network-based heuristic function using backpropagation through steps of the pathfinding process. Our heuristic function learns graph embeddings useful for inferring node distances, runs in constant time independent of graph sizes, and can be easily incorporated in an algorithm such as A* at test time. Experiments show that PHIL reduces the number of explored nodes compared to state-of-the-art methods on benchmark datasets by 58.5\% on average, can be directly applied in diverse graphs ranging from biological networks to road networks, and allows for fast planning in time-critical robotics domains. △ Less

Submitted 10 January, 2023; v1 submitted 7 December, 2022; originally announced December 2022.

DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking

Authors: Gabriele Corso, Hannes Stärk, Bowen **g, Regina Barzilay, Tommi Jaakkola

Abstract: Predicting the binding structure of a small molecule ligand to a protein -- a task known as molecular docking -- is critical to drug design. Recent deep learning methods that treat docking as a regression problem have decreased runtime compared to traditional search-based methods but have yet to offer substantial improvements in accuracy. We instead frame molecular docking as a generative modeling… ▽ More Predicting the binding structure of a small molecule ligand to a protein -- a task known as molecular docking -- is critical to drug design. Recent deep learning methods that treat docking as a regression problem have decreased runtime compared to traditional search-based methods but have yet to offer substantial improvements in accuracy. We instead frame molecular docking as a generative modeling problem and develop DiffDock, a diffusion generative model over the non-Euclidean manifold of ligand poses. To do so, we map this manifold to the product space of the degrees of freedom (translational, rotational, and torsional) involved in docking and develop an efficient diffusion process on this space. Empirically, DiffDock obtains a 38% top-1 success rate (RMSD<2A) on PDBBind, significantly outperforming the previous state-of-the-art of traditional docking (23%) and deep learning (20%) methods. Moreover, while previous methods are not able to dock on computationally folded structures (maximum accuracy 10.4%), DiffDock maintains significantly higher precision (21.7%). Finally, DiffDock has fast inference times and provides confidence estimates with high selective accuracy. △ Less

Submitted 11 February, 2023; v1 submitted 4 October, 2022; originally announced October 2022.

Comments: International Conference on Learning Representations (ICLR 2023)

Torsional Diffusion for Molecular Conformer Generation

Authors: Bowen **g, Gabriele Corso, Jeffrey Chang, Regina Barzilay, Tommi Jaakkola

Abstract: Molecular conformer generation is a fundamental task in computational chemistry. Several machine learning approaches have been developed, but none have outperformed state-of-the-art cheminformatics methods. We propose torsional diffusion, a novel diffusion framework that operates on the space of torsion angles via a diffusion process on the hypertorus and an extrinsic-to-intrinsic score model. On… ▽ More Molecular conformer generation is a fundamental task in computational chemistry. Several machine learning approaches have been developed, but none have outperformed state-of-the-art cheminformatics methods. We propose torsional diffusion, a novel diffusion framework that operates on the space of torsion angles via a diffusion process on the hypertorus and an extrinsic-to-intrinsic score model. On a standard benchmark of drug-like molecules, torsional diffusion generates superior conformer ensembles compared to machine learning and cheminformatics methods in terms of both RMSD and chemical properties, and is orders of magnitude faster than previous diffusion-based models. Moreover, our model provides exact likelihoods, which we employ to build the first generalizable Boltzmann generator. Code is available at https://github.com/gcorso/torsional-diffusion. △ Less

Submitted 28 February, 2023; v1 submitted 1 June, 2022; originally announced June 2022.

Comments: NeurIPS 2022

Acoustic Full Waveform Inversion with Hamiltonian Monte Carlo Method

Authors: Paulo D. S. de Lima, Gilberto Corso, Mauro S. Ferreira, João M. de Araújo

Abstract: Full-Waveform Inversion (FWI) is a high-resolution technique used in geophysics to evaluate the physical parameters and construct subsurface models in a noisy and limited data scenario. The ill-posed nature of the FWI turns this a challenging problem since more than one model can match the observations. In a probabilistic way, solving the FWI problem demands efficient sampling techniques to infer… ▽ More Full-Waveform Inversion (FWI) is a high-resolution technique used in geophysics to evaluate the physical parameters and construct subsurface models in a noisy and limited data scenario. The ill-posed nature of the FWI turns this a challenging problem since more than one model can match the observations. In a probabilistic way, solving the FWI problem demands efficient sampling techniques to infer information on parameters and to estimate the uncertainties in high-dimensional model spaces. We investigate the feasibility of applying the Hamiltonian Monte Carlo (HMC) method in the acoustic FWI by a reflection setup containing different noise level data. We propose a new strategy for tuning the mass matrix based on the acquisition geometry of the seismic survey. Our methodology significantly improves the ability of the HMC method in reconstructing reasonable seismic models with affordable computational efforts. △ Less

Submitted 31 January, 2023; v1 submitted 1 June, 2022; originally announced June 2022.

Comments: 10 pages, 6 figures

Subspace Diffusion Generative Models

Authors: Bowen **g, Gabriele Corso, Renato Berlinghieri, Tommi Jaakkola

Abstract: Score-based models generate samples by map** noise to data (and vice versa) via a high-dimensional diffusion process. We question whether it is necessary to run this entire process at high dimensionality and incur all the inconveniences thereof. Instead, we restrict the diffusion via projections onto subspaces as the data distribution evolves toward noise. When applied to state-of-the-art models… ▽ More Score-based models generate samples by map** noise to data (and vice versa) via a high-dimensional diffusion process. We question whether it is necessary to run this entire process at high dimensionality and incur all the inconveniences thereof. Instead, we restrict the diffusion via projections onto subspaces as the data distribution evolves toward noise. When applied to state-of-the-art models, our framework simultaneously improves sample quality -- reaching an FID of 2.17 on unconditional CIFAR-10 -- and reduces the computational cost of inference for the same number of denoising steps. Our framework is fully compatible with continuous-time diffusion and retains its flexible capabilities, including exact log-likelihoods and controllable generation. Code is available at https://github.com/b**g2016/subspace-diffusion. △ Less

Submitted 27 February, 2023; v1 submitted 3 May, 2022; originally announced May 2022.

Comments: ECCV 2022

Graph Anisotropic Diffusion

Authors: Ahmed A. A. Elhag, Gabriele Corso, Hannes Stärk, Michael M. Bronstein

Abstract: Traditional Graph Neural Networks (GNNs) rely on message passing, which amounts to permutation-invariant local aggregation of neighbour features. Such a process is isotropic and there is no notion of `direction' on the graph. We present a new GNN architecture called Graph Anisotropic Diffusion. Our model alternates between linear diffusion, for which a closed-form solution is available, and local… ▽ More Traditional Graph Neural Networks (GNNs) rely on message passing, which amounts to permutation-invariant local aggregation of neighbour features. Such a process is isotropic and there is no notion of `direction' on the graph. We present a new GNN architecture called Graph Anisotropic Diffusion. Our model alternates between linear diffusion, for which a closed-form solution is available, and local anisotropic filters to obtain efficient multi-hop anisotropic kernels. We test our model on two common molecular property prediction benchmarks (ZINC and QM9) and show its competitive performance. △ Less

Submitted 30 April, 2022; originally announced May 2022.

Comments: 10 pages, 3 figures, Published at the GTRL and MLDD workshops, ICLR 2022

Generalized statistics: applications to data inverse problems with outlier-resistance

Authors: João V. T. de Lima, Sérgio Luiz E. F. da Silva, João M. de Araújo, Gilberto Corso, Gustavo Z. dos Santos Lima

Abstract: The conventional approach to data-driven inversion framework is based on Gaussian statistics that presents serious difficulties, especially in the presence of outliers in the measurements. In this work, we present maximum likelihood estimators associated with generalized Gaussian distributions in the context of Rényi, Tsallis and Kaniadakis statistics. In this regard, we analytically analyse the o… ▽ More The conventional approach to data-driven inversion framework is based on Gaussian statistics that presents serious difficulties, especially in the presence of outliers in the measurements. In this work, we present maximum likelihood estimators associated with generalized Gaussian distributions in the context of Rényi, Tsallis and Kaniadakis statistics. In this regard, we analytically analyse the outlier-resistance of each proposal through the so-called influence function. In this way, we formulate inverse problems by constructing objective functions linked to the maximum likelihood estimators. To demonstrate the robustness of the generalized methodologies, we consider an important geophysical inverse problem with high noisy data with spikes. The results reveal that the best data inversion performance occurs when the entropic index from each generalized statistic is associated with objective functions proportional to the inverse of the error amplitude. We argue that in such a limit the three approaches are resistant to outliers and are also equivalent, which suggests a lower computational cost for the inversion process due to the reduction of numerical simulations to be performed and the fast convergence of the optimization process. △ Less

Submitted 28 January, 2022; originally announced January 2022.

Comments: 17 pages, 14 figures

A fast computational model for the electrophysiology of the whole human heart

Authors: Giulio Del Corso, Roberto Verzicco, Francesco Viola

Abstract: In this study we present a novel computational model for unprecedented simulations of the whole cardiac electrophysiology. According to the heterogeneous electrophysiologic properties of the heart, the whole cardiac geometry is decomposed into a set of coupled conductive media having different topology and electrical conductivities: (i) a network of slender bundles comprising a fast conduction atr… ▽ More In this study we present a novel computational model for unprecedented simulations of the whole cardiac electrophysiology. According to the heterogeneous electrophysiologic properties of the heart, the whole cardiac geometry is decomposed into a set of coupled conductive media having different topology and electrical conductivities: (i) a network of slender bundles comprising a fast conduction atrial network, the AV-node and the ventricular bundles; (ii) the Purkinje network; and (iii) the atrial and ventricular myocardium. The propagation of the action potential in these conductive media is governed by the bidomain/monodomain equations, which are discretized in space using an in-house finite volume method and coupled to three different cellular models, the Courtemanche model [1] for the atrial myocytes, the Stewart model [2] for the Purkinje Network and the ten Tusscher-Panfilov model [3] for the ventricular myocytes. The developed numerical model correctly reproduces the cardiac electrophysiology of the whole human heart in healthy and pathologic conditions and it can be tailored to study and optimize resynchronization therapies or invasive surgical procedures. Importantly, the whole solver is GPU-accelerated using CUDA Fortran providing an unprecedented speedup, thus opening the way for systematic parametric studies and uncertainty quantification analyses. △ Less

Submitted 23 December, 2021; originally announced December 2021.

An outlier-resistant $κ$-generalized approach for robust physical parameter estimation

Authors: Sérgio Luiz E. F. da Silva, R. Silva, Gustavo Z. dos Santos Lima, João M. de Araújo, Gilberto Corso

Abstract: In this work we propose a robust methodology to mitigate the undesirable effects caused by outliers to generate reliable physical models. In this way, we formulate the inverse problems theory in the context of Kaniadakis statistical mechanics (or $κ$-statistics), in which the classical approach is a particular case. In this regard, the errors are assumed to be distributed according to a finite-var… ▽ More In this work we propose a robust methodology to mitigate the undesirable effects caused by outliers to generate reliable physical models. In this way, we formulate the inverse problems theory in the context of Kaniadakis statistical mechanics (or $κ$-statistics), in which the classical approach is a particular case. In this regard, the errors are assumed to be distributed according to a finite-variance $κ$-generalized Gaussian distribution. Based on the probabilistic maximum-likelihood method we derive a $κ$-objective function associated with the finite-variance $κ$-Gaussian distribution. To demonstrate our proposal's outlier-resistance, we analyze the robustness properties of the $κ$-objective function with help of the so-called influence function. In this regard, we discuss the role of the entropic index ($κ$) associated with the Kaniadakis $κ$-entropy in the effectiveness in inferring physical parameters by using strongly noisy data. In this way, we consider a classical geophysical data-inverse problem in two realistic circumstances, in which the first one refers to study the sensibility of our proposal to uncertainties in the input parameters, and the second is devoted to the inversion of a seismic data set contaminated by outliers. The results reveal an optimum $κ$-value at the limit $κ\rightarrow 2/3$, which is related to the best results. △ Less

Submitted 18 November, 2021; originally announced November 2021.

Comments: 37 pages, 23 figures

3D Infomax improves GNNs for Molecular Property Prediction

Authors: Hannes Stärk, Dominique Beaini, Gabriele Corso, Prudencio Tossou, Christian Dallago, Stephan Günnemann, Pietro Liò

Abstract: Molecular property prediction is one of the fastest-growing applications of deep learning with critical real-world impacts. Including 3D molecular structure as input to learned models improves their performance for many molecular tasks. However, this information is infeasible to compute at the scale required by several real-world applications. We propose pre-training a model to reason about the ge… ▽ More Molecular property prediction is one of the fastest-growing applications of deep learning with critical real-world impacts. Including 3D molecular structure as input to learned models improves their performance for many molecular tasks. However, this information is infeasible to compute at the scale required by several real-world applications. We propose pre-training a model to reason about the geometry of molecules given only their 2D molecular graphs. Using methods from self-supervised learning, we maximize the mutual information between 3D summary vectors and the representations of a Graph Neural Network (GNN) such that they contain latent 3D information. During fine-tuning on molecules with unknown geometry, the GNN still generates implicit 3D information and can use it to improve downstream tasks. We show that 3D pre-training provides significant improvements for a wide range of properties, such as a 22% average MAE reduction on eight quantum mechanical properties. Moreover, the learned representations can be effectively transferred between datasets in different molecular spaces. △ Less

Submitted 4 June, 2022; v1 submitted 8 October, 2021; originally announced October 2021.

Comments: 39th International Conference on Machine Learning (ICML 2022). Also accepted at NeurIPS 2021 ML4PH, AI4S, and SSL workshops and as oral at ELLIS ML4Molecules. 24 pages, 7 figures, 18 tables

Journal ref: 39th International Conference on Machine Learning (ICML 2022)

Neural Distance Embeddings for Biological Sequences

Authors: Gabriele Corso, Rex Ying, Michal Pándy, Petar Veličković, Jure Leskovec, Pietro Liò

Abstract: The development of data-dependent heuristics and representations for biological sequences that reflect their evolutionary distance is critical for large-scale biological research. However, popular machine learning approaches, based on continuous Euclidean spaces, have struggled with the discrete combinatorial formulation of the edit distance that models evolution and the hierarchical relationship… ▽ More The development of data-dependent heuristics and representations for biological sequences that reflect their evolutionary distance is critical for large-scale biological research. However, popular machine learning approaches, based on continuous Euclidean spaces, have struggled with the discrete combinatorial formulation of the edit distance that models evolution and the hierarchical relationship that characterises real-world datasets. We present Neural Distance Embeddings (NeuroSEED), a general framework to embed sequences in geometric vector spaces, and illustrate the effectiveness of the hyperbolic space that captures the hierarchical structure and provides an average 22% reduction in embedding RMSE against the best competing geometry. The capacity of the framework and the significance of these improvements are then demonstrated devising supervised and unsupervised NeuroSEED approaches to multiple core tasks in bioinformatics. Benchmarked with common baselines, the proposed approaches display significant accuracy and/or runtime improvements on real-world datasets. As an example for hierarchical clustering, the proposed pretrained and from-scratch methods match the quality of competing baselines with 30x and 15x runtime reduction, respectively. △ Less

Submitted 11 October, 2021; v1 submitted 20 September, 2021; originally announced September 2021.

Comments: Advances in Neural Information Processing Systems (NeurIPS 2021)

ICLR 2021 Challenge for Computational Geometry & Topology: Design and Results

Authors: Nina Miolane, Matteo Caorsi, Umberto Lupo, Marius Guerard, Nicolas Guigui, Johan Mathe, Yann Cabanes, Wojciech Reise, Thomas Davies, António Leitão, Somesh Mohapatra, Saiteja Utpala, Shailja Shailja, Gabriele Corso, Guoxi Liu, Federico Iuricich, Andrei Manolache, Mihaela Nistor, Matei Bejan, Armand Mihai Nicolicioiu, Bogdan-Alexandru Luchian, Mihai-Sorin Stupariu, Florent Michel, Khanh Dao Duc, Bilal Abdulrahman , et al. (8 additional authors not shown)

Abstract: This paper presents the computational challenge on differential geometry and topology that happened within the ICLR 2021 workshop "Geometric and Topological Representation Learning". The competition asked participants to provide creative contributions to the fields of computational geometry and topology through the open-source repositories Geomstats and Giotto-TDA. The challenge attracted 16 teams… ▽ More This paper presents the computational challenge on differential geometry and topology that happened within the ICLR 2021 workshop "Geometric and Topological Representation Learning". The competition asked participants to provide creative contributions to the fields of computational geometry and topology through the open-source repositories Geomstats and Giotto-TDA. The challenge attracted 16 teams in its two month duration. This paper describes the design of the challenge and summarizes its main findings. △ Less

Submitted 25 August, 2021; v1 submitted 22 August, 2021; originally announced August 2021.

Orthogonal iterations on Structured Pencils

Authors: Roberto Bevilacqua, Gianna M. Del Corso, Luca Gemignani

Abstract: We present a class of fast subspace tracking algorithms based on orthogonal iterations for structured matrices/pencils that can be represented as small rank perturbations of unitary matrices. The algorithms rely upon an updated data sparse factorization -- named LFR factorization -- using orthogonal Hessenberg matrices. These new subspace trackers reach a complexity of only $O(nk^2)$ operations pe… ▽ More We present a class of fast subspace tracking algorithms based on orthogonal iterations for structured matrices/pencils that can be represented as small rank perturbations of unitary matrices. The algorithms rely upon an updated data sparse factorization -- named LFR factorization -- using orthogonal Hessenberg matrices. These new subspace trackers reach a complexity of only $O(nk^2)$ operations per time update, where $n$ and $k$ are the size of the matrix and of the small rank perturbation, respectively. △ Less

Submitted 22 April, 2021; originally announced April 2021.

MSC Class: 65F15 ACM Class: G.1.3

Parameter free determination of optimum time delay

Authors: Thiago Lima Prado, Vandertone Santos Machado, Gilberto Corso, Gustavo Zampier dos Santos Lima, Sergio Roberto Lopes

Abstract: We show that the same maximum entropy principle applied to recurrence microstates configures a new way to properly compute the time delay necessary to correctly sample a data set. The new method retrieves results obtained using traditional methods with the advantage of being independent of any free parameter. Since all parameters are automatically set, the method is suitable for use in artificial… ▽ More We show that the same maximum entropy principle applied to recurrence microstates configures a new way to properly compute the time delay necessary to correctly sample a data set. The new method retrieves results obtained using traditional methods with the advantage of being independent of any free parameter. Since all parameters are automatically set, the method is suitable for use in artificial (computational) intelligence algorithms, recovering correct information embedded in time series, and rationalizing the process of data acquisition since only relevant data must be collected. △ Less

Submitted 6 October, 2020; originally announced October 2020.

Comments: 5 pages, 4 figures

Directional Graph Networks

Authors: Dominique Beaini, Saro Passaro, Vincent Létourneau, William L. Hamilton, Gabriele Corso, Pietro Liò

Abstract: The lack of anisotropic kernels in graph neural networks (GNNs) strongly limits their expressiveness, contributing to well-known issues such as over-smoothing. To overcome this limitation, we propose the first globally consistent anisotropic kernels for GNNs, allowing for graph convolutions that are defined according to topologicaly-derived directional flows. First, by defining a vector field in t… ▽ More The lack of anisotropic kernels in graph neural networks (GNNs) strongly limits their expressiveness, contributing to well-known issues such as over-smoothing. To overcome this limitation, we propose the first globally consistent anisotropic kernels for GNNs, allowing for graph convolutions that are defined according to topologicaly-derived directional flows. First, by defining a vector field in the graph, we develop a method of applying directional derivatives and smoothing by projecting node-specific messages into the field. Then, we propose the use of the Laplacian eigenvectors as such vector field. We show that the method generalizes CNNs on an $n$-dimensional grid and is provably more discriminative than standard GNNs regarding the Weisfeiler-Lehman 1-WL test. We evaluate our method on different standard benchmarks and see a relative error reduction of 8% on the CIFAR10 graph dataset and 11% to 32% on the molecular ZINC dataset, and a relative increase in precision of 1.6% on the MolPCBA dataset. An important outcome of this work is that it enables graph networks to embed directions in an unsupervised way, thus allowing a better representation of the anisotropic features in different physical or biological problems. △ Less

Submitted 7 April, 2021; v1 submitted 6 October, 2020; originally announced October 2020.

Comments: 11 pages, 10 pages appendix, 6 figures, subtitle: Anisotropic aggregation in graph neural networks via directional vector fields

Principal Neighbourhood Aggregation for Graph Nets

Authors: Gabriele Corso, Luca Cavalleri, Dominique Beaini, Pietro Liò, Petar Veličković

Abstract: Graph Neural Networks (GNNs) have been shown to be effective models for different predictive tasks on graph-structured data. Recent work on their expressive power has focused on isomorphism tasks and countable feature spaces. We extend this theoretical framework to include continuous features - which occur regularly in real-world input domains and within the hidden layers of GNNs - and we demonstr… ▽ More Graph Neural Networks (GNNs) have been shown to be effective models for different predictive tasks on graph-structured data. Recent work on their expressive power has focused on isomorphism tasks and countable feature spaces. We extend this theoretical framework to include continuous features - which occur regularly in real-world input domains and within the hidden layers of GNNs - and we demonstrate the requirement for multiple aggregation functions in this context. Accordingly, we propose Principal Neighbourhood Aggregation (PNA), a novel architecture combining multiple aggregators with degree-scalers (which generalize the sum aggregator). Finally, we compare the capacity of different models to capture and exploit the graph structure via a novel benchmark containing multiple tasks taken from classical graph theory, alongside existing benchmarks from real-world domains, all of which demonstrate the strength of our model. With this work, we hope to steer some of the GNN research towards new aggregation methods which we believe are essential in the search for powerful and robust models. △ Less

Submitted 31 December, 2020; v1 submitted 12 April, 2020; originally announced April 2020.

Comments: 34th Conference on Neural Information Processing Systems (NeurIPS 2020)

Parameter-free quantification of stochastic and chaotic signals

Authors: Sergio Roberto Lopes, Thiago de Lima Prado, Gilberto Corso, Gustavo Zampier dos Santos Lima, Jurgen Kurths

Abstract: Recurrence entropy $(\cal S)$ is a novel time series complexity quantifier based on recurrence microstates. Here we show that $\mathsf{max}(\cal S)$ is a \textit{parameter-free} quantifier of time correlation of stochastic and chaotic signals, at the same time that it evaluates property changes of the probability distribution function (PDF) of the entire data set. $\mathsf{max}(\cal S)$ can distin… ▽ More Recurrence entropy $(\cal S)$ is a novel time series complexity quantifier based on recurrence microstates. Here we show that $\mathsf{max}(\cal S)$ is a \textit{parameter-free} quantifier of time correlation of stochastic and chaotic signals, at the same time that it evaluates property changes of the probability distribution function (PDF) of the entire data set. $\mathsf{max}(\cal S)$ can distinguish distinct temporal correlations of stochastic signals following a power-law spectrum, $\displaystyle P(f) \propto 1/f^α$ even when shuffled versions of the signals are used. Such behavior is related to its ability to quantify distinct subsets embedded in a time series. Applied to a deterministic system, the method brings new evidence about attractor properties and the degree of chaoticity. The development of a new parameter-free quantifier of stochastic and chaotic time series opens new perspectives to stochastic data and deterministic time series analyses and may find applications in many areas of science. △ Less

Submitted 6 May, 2019; originally announced May 2019.

Efficient Reduction of Compressed Unitary plus Low-rank Matrices to Hessenberg form

Authors: Roberto Bevilacqua, Gianna M. Del Corso, Luca Gemignani

Abstract: We present fast numerical methods for computing the Hessenberg reduction of a unitary plus low-rank matrix $A=G+U V^H$, where $G\in \mathbb C^{n\times n}$ is a unitary matrix represented in some compressed format using $O(nk)$ parameters and $U$ and $V$ are $n\times k$ matrices with $k< n$. At the core of these methods is a certain structured decomposition, referred to as a LFR decomposition, of… ▽ More We present fast numerical methods for computing the Hessenberg reduction of a unitary plus low-rank matrix $A=G+U V^H$, where $G\in \mathbb C^{n\times n}$ is a unitary matrix represented in some compressed format using $O(nk)$ parameters and $U$ and $V$ are $n\times k$ matrices with $k< n$. At the core of these methods is a certain structured decomposition, referred to as a LFR decomposition, of $A$ as product of three possibly perturbed unitary $k$ Hessenberg matrices of size $n$. It is shown that in most interesting cases an initial LFR decomposition of $A$ can be computed very cheaply. Then we prove structural properties of LFR decompositions by giving conditions under which the LFR decomposition of $A$ implies its Hessenberg shape. Finally, we describe a bulge chasing scheme for converting the initial LFR decomposition of $A$ into the LFR decomposition of a Hessenberg matrix by means of unitary transformations. The reduction can be performed at the overall computational cost of $O(n^2 k)$ arithmetic operations using $O(nk)$ storage. The computed LFR decomposition of the Hessenberg reduction of $A$ can be processed by the fast QR algorithm presented in [8] in order to compute the eigenvalues of $A$ within the same costs. △ Less

Submitted 29 August, 2019; v1 submitted 24 January, 2019; originally announced January 2019.

MSC Class: 65F15;

When is a matrix unitary or Hermitian plus low rank?

Authors: Gianna M. Del Corso, Federico Poloni, Leonardo Robol, Raf Vandebril

Abstract: Hermitian and unitary matrices are two representatives of the class of normal matrices whose full eigenvalue decomposition can be stably computed in quadratic computing com plexity. Recently, fast and reliable eigensolvers dealing with low rank perturbations of unitary and Hermitian matrices were proposed. These structured eigenvalue problems appear naturally when computing roots, via confederate… ▽ More Hermitian and unitary matrices are two representatives of the class of normal matrices whose full eigenvalue decomposition can be stably computed in quadratic computing com plexity. Recently, fast and reliable eigensolvers dealing with low rank perturbations of unitary and Hermitian matrices were proposed. These structured eigenvalue problems appear naturally when computing roots, via confederate linearizations, of polynomials expressed in, e.g., the monomial or Chebyshev basis. Often, however, it is not known beforehand whether or not a matrix can be written as the sum of an Hermitian or unitary matrix plus a low rank perturbation. We propose necessary and sufficient conditions characterizing the class of Hermitian or unitary plus low rank matrices. The number of singular values deviating from 1 determines the rank of a perturbation to bring a matrix to unitary form. A similar condition holds for Hermitian matrices; the eigenvalues of the skew-Hermitian part differing from 0 dictate the rank of the perturbation. We prove that these relations are linked via the Cayley transform. Based on these conditions we are able to identify the closest Hermitian and unitary plus low rank matrix in Frobenius and spectral norm and a practical Lanczos iteration to detect the low rank perturbation is presented. Numerical tests prove that this straightforward algorithm is robust with respect to noise. △ Less

Submitted 25 July, 2019; v1 submitted 14 November, 2018; originally announced November 2018.

Comments: (to appear)

MSC Class: 15B10; 15B57; 65FXX

Journal ref: Numerical Linear Algebra with Applications, 2019

Fast QR iterations for unitary plus low rank matrices

Authors: Roberto Bevilacqua, Gianna M. Del Corso, Luca Gemignani

Abstract: Some fast algorithms for computing the eigenvalues of a block companion matrix $A = U + XY^H$, where $U\in \mathbb C^{n\times n}$ is unitary block circulant and $X, Y \in\mathbb{C}^{n \times k}$, have recently appeared in the literature. Most of these algorithms rely on the decomposition of $A$ as product of scalar companion matrices which turns into a factored representation of the Hessenberg red… ▽ More Some fast algorithms for computing the eigenvalues of a block companion matrix $A = U + XY^H$, where $U\in \mathbb C^{n\times n}$ is unitary block circulant and $X, Y \in\mathbb{C}^{n \times k}$, have recently appeared in the literature. Most of these algorithms rely on the decomposition of $A$ as product of scalar companion matrices which turns into a factored representation of the Hessenberg reduction of $A$. In this paper we generalize the approach to encompass Hessenberg matrices of the form $A=U + XY^H$ where $U$ is a general unitary matrix. A remarkable case is $U$ unitary diagonal which makes possible to deal with interpolation techniques for rootfinding problems and nonlinear eigenvalue problems. Our extension exploits the properties of a larger matrix $\hat A$ obtained by a certain embedding of the Hessenberg reduction of $A$ suitable to maintain its structural properties. We show that $\hat A$ can be factored as product of lower and upper unitary Hessenberg matrices possibly perturbed in the first $k$ rows, and, moreover, such a data-sparse representation is well suited for the design of fast eigensolvers based on the QR/QZ iteration. The resulting algorithm is fast and backward stable. △ Less

Submitted 29 August, 2019; v1 submitted 5 October, 2018; originally announced October 2018.

MSC Class: 65F15

A novel entropy recurrence quantification analysis

Authors: G. Corso, T. L. Prado, G. Z. dos S. Lima, S. R. Lopes

Abstract: The growing study of time series, especially those related to nonlinear systems, has challenged the methodologies to characterize and classify dynamical structures of a signal. Here we conceive a new diagnostic tool for time series based on the concept of information entropy, in which the probabilities are associated to microstates defined from the recurrence phase space. Recurrence properties can… ▽ More The growing study of time series, especially those related to nonlinear systems, has challenged the methodologies to characterize and classify dynamical structures of a signal. Here we conceive a new diagnostic tool for time series based on the concept of information entropy, in which the probabilities are associated to microstates defined from the recurrence phase space. Recurrence properties can properly be studied using recurrence plots, a methodology based on binary matrices where trajec- tories in phase space of dynamical systems are evaluated against other embedded trajectory. Our novel entropy methodology has several advantages compared to the traditional recurrence entropy defined in the literature, namely, the correct evaluation of the chaoticity level of the signal, the weak dependence on parameters, correct evaluation of periodic time series properties and more sensitivity to noise level of time series. Furthermore, the new entropy quantifier developed in this manuscript also fixes inconsistent results of the traditional recurrence entropy concept, reproducing classical results with novel insights. △ Less

Submitted 4 July, 2017; originally announced July 2017.

Universality and the collapse of multifractality in Barkhausen avalanches

Authors: Gustavo Zampier dos Santos Lima, Gilberto Corso, Marcio Assolin Corrêa, Rubem Luis Sommer, Plamen Ch. Ivanov, Felipe Bohn

Abstract: Barkhausen effect in ferromagnetic materials provides an excellent area for investigating scaling phenomena found in disordered systems exhibiting crackling noise. The critical dynamics is characterized by random pulses or avalanches with scale-invariant properties, power-law distributions, and universal features. However, the traditional Barkhausen avalanches statistics may not be sufficient to f… ▽ More Barkhausen effect in ferromagnetic materials provides an excellent area for investigating scaling phenomena found in disordered systems exhibiting crackling noise. The critical dynamics is characterized by random pulses or avalanches with scale-invariant properties, power-law distributions, and universal features. However, the traditional Barkhausen avalanches statistics may not be sufficient to fully characterize the complex temporal correlation of the magnetic domain walls dynamics. Here we go beyond power laws and focus on the multifractal scenario to quantify the temporal scaling characteristics of Barkhausen avalanches in polycrystalline and amorphous ferromagnetic films with thicknesses from $50$ nm to $1000$ nm. We show that the multifractal properties are dependent on the film thickness, and insensitive to the structural character of the materials. Further, we observe for the first time the collapse of the multifractality in the domain walls dynamics as the thickness is reduced, and the multifractal behavior gives place to a monofractal one over the entire range of time scales. The reorganization in the temporal scaling characteristics of Barkhausen avalanches is understood as an universal restructuring associated to the dimensional transition, from three to two-dimensional magnetization dynamics. △ Less

Submitted 4 November, 2016; originally announced November 2016.

Comments: 3 figures

Journal ref: Phys. Rev. E 96, 022159 (2017)

Adaptive Nonnegative Matrix Factorization and Measure Comparisons for Recommender Systems

Authors: Gianna M. Del Corso, Francesco Romani

Abstract: The Nonnegative Matrix Factorization (NMF) of the rating matrix has shown to be an effective method to tackle the recommendation problem. In this paper we propose new methods based on the NMF of the rating matrix and we compare them with some classical algorithms such as the SVD and the regularized and unregularized non-negative matrix factorization approach. In particular a new algorithm is obtai… ▽ More The Nonnegative Matrix Factorization (NMF) of the rating matrix has shown to be an effective method to tackle the recommendation problem. In this paper we propose new methods based on the NMF of the rating matrix and we compare them with some classical algorithms such as the SVD and the regularized and unregularized non-negative matrix factorization approach. In particular a new algorithm is obtained changing adaptively the function to be minimized at each step, realizing a sort of dynamic prior strategy. Another algorithm is obtained modifying the function to be minimized in the NMF formulation by enforcing the reconstruction of the unknown ratings toward a prior term. We then combine different methods obtaining two mixed strategies which turn out to be very effective in the reconstruction of missing observations. We perform a thoughtful comparison of different methods on the basis of several evaluation measures. We consider in particular rating, classification and ranking measures showing that the algorithm obtaining the best score for a given measure is in general the best also when different measures are considered, lowering the interest in designing specific evaluation measures. The algorithms have been tested on different datasets, in particular the 1M, and 10M MovieLens datasets containing ratings on movies, the Jester dataset with ranting on jokes and Amazon Fine Foods dataset with ratings on foods. The comparison of the different algorithms, shows the good performance of methods employing both an explicit and an implicit regularization scheme. Moreover we can get a boost by mixed strategies combining a fast method with a more accurate one. △ Less

Submitted 29 August, 2019; v1 submitted 26 July, 2016; originally announced July 2016.

MSC Class: 65F99

Journal ref: Applied Mathematics and Computation 354, pp. 164-179, 2019

Stationarity breaking in coupled physical systems revealed by recurrence analysis

Authors: Thiago de Lima Prado, Gustavo Zampier dos Santos Lima, Bruno Lobão-Soares, George Carlos do Nascimento, Gilberto Corso, Sergio Roberto Lopes

Abstract: In this letter we explore how recurrence quantifier, the determinism ($Δ$), can reveal stationarity breaking and coupling between physical systems. We demonstrate that it is possible to detect small variations in a dynamical system based only on temporal signal displayed by another system coupled to it. To introduce basic ideas, we consider a well known dynamical system composed of two master-slav… ▽ More In this letter we explore how recurrence quantifier, the determinism ($Δ$), can reveal stationarity breaking and coupling between physical systems. We demonstrate that it is possible to detect small variations in a dynamical system based only on temporal signal displayed by another system coupled to it. To introduce basic ideas, we consider a well known dynamical system composed of two master-slave coupled Lorenz oscillators. We start evidencing that due to the sensitivity of $Δ$ computed from temporal time series of slave oscillator, its is possible to detect the stationary breaking imposed in the master oscillator. As a second example, the method is carried out in a real physiological data acquired from accelerometer sensors ($\mathrm{A_{cc}}$) and used to detect micro arousal phenomenology (described by a sharp burst in $\mathrm{A_{cc}}$ signal) during sleep periods in mice. Moreover, we show for the first time that making use of recurrence quantifier it is possible to infer a coupling between electric signals from hippocampus and "locomotor brain areas" of mice, based only on non invasive data from $\mathrm{A_{cc}}$. Our results suggest new possibilities of analysis of coupled systems making use of accessible time series. Our second example supports an interpretation of an internal coupling detectable as a stationarity breaking in $\mathrm{A_{cc}}$ that occurs some seconds before micro arousal processes during sleep periods in rodents, contributing to the idea that micro arousals are elements of sleep taking part in the regulation of sleep process. Such a characterization of micro arousals can improve our knowledge about sleep fostering tools of sleep diagnose and pharmacology research for mammals in general. △ Less

Submitted 4 April, 2016; v1 submitted 12 February, 2016; originally announced February 2016.

Comments: After the review process, the authors do not agree that the paper should be submitted to the arXv. So no future version of the text will be uploaded

A multi-class approach for ranking graph nodes: models and experiments with incomplete data

Authors: Gianna M. Del Corso, Francesco Romani

Abstract: After the phenomenal success of the PageRank algorithm, many researchers have extended the PageRank approach to ranking graphs with richer structures beside the simple linkage structure. In some scenarios we have to deal with multi-parameters data where each node has additional features and there are relationships between such features. This paper stems from the need of a systematic approach whe… ▽ More After the phenomenal success of the PageRank algorithm, many researchers have extended the PageRank approach to ranking graphs with richer structures beside the simple linkage structure. In some scenarios we have to deal with multi-parameters data where each node has additional features and there are relationships between such features. This paper stems from the need of a systematic approach when dealing with multi-parameter data. We propose models and ranking algorithms which can be used with little adjustments for a large variety of networks (bibliographic data, patent data, twitter and social data, healthcare data). In this paper we focus on several aspects which have not been addressed in the literature: (1) we propose different models for ranking multi-parameters data and a class of numerical algorithms for efficiently computing the ranking score of such models, (2) by analyzing the stability and convergence properties of the numerical schemes we tune a fast and stable technique for the ranking problem, (3) we consider the issue of the robustness of our models when data are incomplete. The comparison of the rank on the incomplete data with the rank on the full structure shows that our models compute consistent rankings whose correlation is up to 60% when just 10% of the links of the attributes are maintained suggesting the suitability of our model also when the data are incomplete. △ Less

Submitted 29 April, 2015; originally announced April 2015.

MSC Class: 65F15 ACM Class: G.2.2; F.2.1

Journal ref: Information Sciences 2016, vol 329 pages 619-637

A CMV--based eigensolver for companion matrices

Authors: Roberto Bevilacqua, Gianna M. Del Corso, Luca gemignani

Abstract: In this paper we present a novel matrix method for polynomial rootfinding. By exploiting the properties of the QR eigenvalue algorithm applied to a suitable CMV-like form of a companion matrix we design a fast and computationally simple structured QR iteration. In this paper we present a novel matrix method for polynomial rootfinding. By exploiting the properties of the QR eigenvalue algorithm applied to a suitable CMV-like form of a companion matrix we design a fast and computationally simple structured QR iteration. △ Less

Submitted 11 June, 2014; originally announced June 2014.

Comments: 14 pages, 4 figures

MSC Class: 65F15

Journal ref: Siam J. Matrix Anal. Appl. 2015 Vol. 36, No. 3, pp. 1046-1068

Compression of unitary rank--structured matrices to CMV-like shape with an application to polynomial rootfinding

Authors: Roberto Bevilacqua, Gianna M. Del Corso, Luca Gemignani

Abstract: This paper is concerned with the reduction of a unitary matrix U to CMV-like shape. A Lanczos--type algorithm is presented which carries out the reduction by computing the block tridiagonal form of the Hermitian part of U, i.e., of the matrix U+U^H. By elaborating on the Lanczos approach we also propose an alternative algorithm using elementary matrices which is numerically stable. If U is rank--s… ▽ More This paper is concerned with the reduction of a unitary matrix U to CMV-like shape. A Lanczos--type algorithm is presented which carries out the reduction by computing the block tridiagonal form of the Hermitian part of U, i.e., of the matrix U+U^H. By elaborating on the Lanczos approach we also propose an alternative algorithm using elementary matrices which is numerically stable. If U is rank--structured then the same property holds for its Hermitian part and, therefore, the block tridiagonalization process can be performed using the rank--structured matrix technology with reduced complexity. Our interest in the CMV-like reduction is motivated by the unitary and almost unitary eigenvalue problem. In this respect, finally, we discuss the application of the CMV-like reduction for the design of fast companion eigensolvers based on the customary QR iteration. △ Less

Submitted 8 July, 2013; originally announced July 2013.

MSC Class: 65F15

Journal ref: Journal of Computational and Applied Mathematics 2015 vol. 278 326-335

Block Tridiagonal Reduction of Perturbed Normal and Rank Structured Matrices

Authors: Roberto Bevilacqua, Gianna M. Del Corso, Luca Gemignani

Abstract: It is well known that if a matrix $A\in\mathbb C^{n\times n}$ solves the matrix equation $f(A,A^H)=0$, where $f(x, y)$ is a linear bivariate polynomial, then $A$ is normal; $A$ and $A^H$ can be simultaneously reduced in a finite number of operations to tridiagonal form by a unitary congruence and, moreover, the spectrum of $A$ is located on a straight line in the complex plane. In this paper we pr… ▽ More It is well known that if a matrix $A\in\mathbb C^{n\times n}$ solves the matrix equation $f(A,A^H)=0$, where $f(x, y)$ is a linear bivariate polynomial, then $A$ is normal; $A$ and $A^H$ can be simultaneously reduced in a finite number of operations to tridiagonal form by a unitary congruence and, moreover, the spectrum of $A$ is located on a straight line in the complex plane. In this paper we present some generalizations of these properties for almost normal matrices which satisfy certain quadratic matrix equations arising in the study of structured eigenvalue problems for perturbed Hermitian and unitary matrices. △ Less

Submitted 24 June, 2013; originally announced June 2013.

Comments: 13 pages, 3 figures

MSC Class: 65F15

Journal ref: Linear Algebra and Its Applications 439(11), pp. 3505-3517 2013

The infinite partition of a line segment and multifractal objects

Authors: A. I. L. de Araújo, R. F. Soares, J. P. de Oliveira, G. Corso

Abstract: We report an algorithm for the partition of a line segment according to a given ratio $ν$. At each step the length distribution among sets of the partition follows a binomial distribution. We call $k$-set to the set of elements with the same length at the step $n$. The total number of elements is $2^n$ and the number of elements in a same $k$-set is $C_n^k$. In the limit of an infinite partion t… ▽ More We report an algorithm for the partition of a line segment according to a given ratio $ν$. At each step the length distribution among sets of the partition follows a binomial distribution. We call $k$-set to the set of elements with the same length at the step $n$. The total number of elements is $2^n$ and the number of elements in a same $k$-set is $C_n^k$. In the limit of an infinite partion this object become a multifractal where each $k$-set originate a fractal. We find the fractal spectrum $D_k$ and calculate where is its maximum. Finally we find the values of $D_k$ for the limits $k/n \to 0$ and 1. △ Less

Submitted 7 November, 2008; originally announced November 2008.

A new nestedness estimator in community networks

Authors: Gilberto Corso, Aderaldo I. L. Araujo, Adriana M. Almeida

Abstract: A recent problem in community ecology lies in defining structures behind matrices of species interactions. The interest in this area is to quantify the nestedness degree of the matrix after its maximal packing. In this work we evaluate nestedness using the sum of all distances of the occupied sites to the vertex of the matrix. We calculate the distance for two artificial matrices with the same s… ▽ More A recent problem in community ecology lies in defining structures behind matrices of species interactions. The interest in this area is to quantify the nestedness degree of the matrix after its maximal packing. In this work we evaluate nestedness using the sum of all distances of the occupied sites to the vertex of the matrix. We calculate the distance for two artificial matrices with the same size and occupancy: a random matrix and a perfect nested one. Using these two benchmarks we develop a nestedness estimator. The estimator is applied to a set of 23 real networks of insect-plant interactions. △ Less

Submitted 29 February, 2008; originally announced March 2008.

Comments: 16 pages, 8 figures, 1 table

A Fractal Space-filling Complex Network

Authors: D. J. B. Soares, J. Ribeiro Filho, A. A. Moreira, D. A. Moreira, G. Corso

Abstract: We study in this work the properties of the $Q_{mf}$ network which is constructed from an anisotropic partition of the square, the multifractal tiling. This tiling is build using a single parameter $ρ$, in the limit of $ρ\to 1$ the tiling degenerates into the square lattice that is associated with a regular network. The $Q_{mf}$ network is a space-filling network with the following characteris… ▽ More We study in this work the properties of the $Q_{mf}$ network which is constructed from an anisotropic partition of the square, the multifractal tiling. This tiling is build using a single parameter $ρ$, in the limit of $ρ\to 1$ the tiling degenerates into the square lattice that is associated with a regular network. The $Q_{mf}$ network is a space-filling network with the following characteristics: it shows a power-law distribution of connectivity for $k>7$ and it has an high clustering coefficient when compared with a random network associated. In addition the $Q_{mf}$ network satisfy the relation $N \propto \ell^{d_f}$ where $\ell$ is a typical length of the network (the average minimal distance) and $N$ the network size. We call $d_f$ the fractal dimension of the network. In tne limit case $ρ\to 1$ we have $d_{f} \to 2$. △ Less

Submitted 15 August, 2005; originally announced August 2005.

Comments: 10 pages, 5 figures and 1 table

A Random Multifractal Tilling

Authors: M. G. Pereira, G. Corso, L. S. Lucena, J. E. Freitas

Abstract: We develop a multifractal random tilling that fills the square. The multifractal is formed by an arrangement of rectangular blocks of different sizes, areas and number of neighbors. The overall feature of the tilling is an heterogeneous and anisotropic random self-affine object. The multifractal is constructed by an algorithm that makes successive sections of the square. At each $n$-step there i… ▽ More We develop a multifractal random tilling that fills the square. The multifractal is formed by an arrangement of rectangular blocks of different sizes, areas and number of neighbors. The overall feature of the tilling is an heterogeneous and anisotropic random self-affine object. The multifractal is constructed by an algorithm that makes successive sections of the square. At each $n$-step there is a random choice of a parameter $ρ_i$ related to the section ratio. For the case of random choice between $ρ_1$ and $ρ_2$ we find analytically the full spectrum of fractal dimensions. △ Less

Submitted 13 February, 2004; originally announced February 2004.

Varying critical percolation exponents on a multifractal support

Authors: J. E. Freitas, G. Corso, L. S. Lucena

Abstract: We study percolation as a critical phenomenon on a multifractal support. The scaling exponents of the the infinite cluster size ($β$ exponent) and the fractal dimension of the percolation cluster ($d_f$) are quantities that seem do not depend on local anisotropies. These two quantities have the same value as in the standard percolation in regular bidimensional lattices. On the other side, the sc… ▽ More We study percolation as a critical phenomenon on a multifractal support. The scaling exponents of the the infinite cluster size ($β$ exponent) and the fractal dimension of the percolation cluster ($d_f$) are quantities that seem do not depend on local anisotropies. These two quantities have the same value as in the standard percolation in regular bidimensional lattices. On the other side, the scaling of the correlation length ($ν$ exponent) unfolds new universality classes due to the local anisotropy of the critical percolation cluster. We use two critical exponents $ν$ according to the percolation criterion for crossing the lattice in either direction or in both directions. Moreover $ν$ is related to a parameter that characterizes the stretching of the blocks forming the tilling of the multifractal. △ Less

Submitted 31 October, 2003; originally announced October 2003.

Families and clustering in a natural numbers network

Authors: Gilberto Corso

Abstract: We develop a network in which the natural numbers are the vertices. We use the decomposition of natural numbers by prime numbers to establish the connections. We perform data collapse and show that the degree distribution of these networks scale linearly with the number of vertices. We compare the average distance of the network and the clustering coefficient with the distance and clustering coe… ▽ More We develop a network in which the natural numbers are the vertices. We use the decomposition of natural numbers by prime numbers to establish the connections. We perform data collapse and show that the degree distribution of these networks scale linearly with the number of vertices. We compare the average distance of the network and the clustering coefficient with the distance and clustering coefficient of the corresponding random graph. In case we set connections among vertices each time the numbers share a common prime number the network is not a small-world type. If the criterium for establishing links becomes more selective, only prime numbers greater than $p_l$ are used to establish links, the network shows small-world effect, it means, it has high clustering coefficient and low distance. △ Less

Submitted 24 November, 2003; v1 submitted 8 September, 2003; originally announced September 2003.

Anisotropy and percolation threshold in a multifractal support

Authors: L. S. Lucena, J. E. Freitas, G. Corso, R. F. Soares

Abstract: Recently a multifractal object, $Q_{mf}$, was proposed to study percolation properties in a multifractal support. The area and the number of neighbors of the blocks of $Q_{mf}$ show a non-trivial behavior. The value of the probability of occupation at the percolation threshold, $p_{c}$, is a function of $ρ$, a parameter of $Q_{mf}$ which is related to its anisotropy. We investigate the relation… ▽ More Recently a multifractal object, $Q_{mf}$, was proposed to study percolation properties in a multifractal support. The area and the number of neighbors of the blocks of $Q_{mf}$ show a non-trivial behavior. The value of the probability of occupation at the percolation threshold, $p_{c}$, is a function of $ρ$, a parameter of $Q_{mf}$ which is related to its anisotropy. We investigate the relation between $p_{c}$ and the average number of neighbors of the blocks as well as the anisotropy of $Q_{mf}$. △ Less

Submitted 14 August, 2003; originally announced August 2003.

Percolation in a Multifractal

Authors: G. Corso, J. E. Freitas, L. S. Lucena, R. F. Soares

Abstract: We build a multifractal object and use it as a support to study percolation. We identify some differences between percolation in a multifractal and in a regular lattice. We use many samples of finite size lattices and draw the histogram of percolating lattices against site occupation probability. Depending on a parameter characterizing the multifractal and the lattice size, the histogram can ha… ▽ More We build a multifractal object and use it as a support to study percolation. We identify some differences between percolation in a multifractal and in a regular lattice. We use many samples of finite size lattices and draw the histogram of percolating lattices against site occupation probability. Depending on a parameter characterizing the multifractal and the lattice size, the histogram can have two peaks. The percolation threshold for the multifractal is lower than for the square lattice. The percolation in the multifractal differs from the percolation in the regular lattice in two points. The first is related with the coordination number that changes along the multifractal. The second comes from the way the weight of each cell in the multifractal affects the percolation cluster. We compute the fractal dimension of the percolating cluster. Despite the differences, the percolation in a multifractal support is in the universality class of standard percolation. △ Less

Submitted 11 August, 2003; v1 submitted 20 December, 2002; originally announced December 2002.

Mechanical Mixing in Nonlinear Nanomechanical Resonators

Authors: A. Erbe, G. Corso, H. Krommer, A. Kraus, K. Richter, R. H. Blick

Abstract: Nanomechanical resonators, machined out of Silicon-on-Insulator wafers, are operated in the nonlinear regime to investigate higher-order mechanical mixing at radio frequencies, relevant to signal processing and nonlinear dynamics on nanometer scales. Driven by two neighboring frequencies the resonators generate rich power spectra exhibiting a multitude of satellite peaks. This nonlinear response… ▽ More Nanomechanical resonators, machined out of Silicon-on-Insulator wafers, are operated in the nonlinear regime to investigate higher-order mechanical mixing at radio frequencies, relevant to signal processing and nonlinear dynamics on nanometer scales. Driven by two neighboring frequencies the resonators generate rich power spectra exhibiting a multitude of satellite peaks. This nonlinear response is studied and compared to $n^{th}$-order perturbation theory and nonperturbative numerical calculations. △ Less

Submitted 15 December, 1999; originally announced December 1999.

Comments: 5 pages, 7 figures

Report number: mpi-pks/9912003

Separatrix Reconnections in Chaotic Regimes

Authors: G. Corso, F. B. Rizzato

Abstract: In this paper we extend the concept of separatrix reconnection into chaotic regimes. We show that even under chaotic conditions one can still understand abrupt jumps of diffusive-like processes in the relevant phase-space in terms of relatively smooth realignments of stable and unstable manifolds of unstable fixed points. In this paper we extend the concept of separatrix reconnection into chaotic regimes. We show that even under chaotic conditions one can still understand abrupt jumps of diffusive-like processes in the relevant phase-space in terms of relatively smooth realignments of stable and unstable manifolds of unstable fixed points. △ Less

Submitted 15 January, 1998; v1 submitted 8 January, 1998; originally announced January 1998.

Comments: 4 pages, 5 figures, submitted do Phys. Rev. E (1998)

Report number: IF-RIZa-0198