Search

We introduce a novel modeling approach for time series imputation and forecasting, tailored to address the challenges often encountered in real-world data, such as irregular samples, missing data, or unaligned measurements from multiple sensors. Our method relies on a continuous-time-dependent model of the series' evolution dynamics. It leverages adaptations of conditional, implicit neural representations for sequential data. A modulation mechanism, driven by a meta-learning algorithm, allows adaptation to unseen samples and extrapolation beyond observed time-windows for long-term predictions. The model provides a highly flexible and unified framework for imputation and forecasting tasks across a wide range of challenging scenarios. It achieves state-of-the-art performance on classical benchmarks and outperforms alternative time-continuous models.

Two recent developments have accelerated progress in image reconstruction from human brain activity: large datasets that offer samples of brain activity in response to many thousands of natural scenes, and the open-sourcing of powerful stochastic image-generators that accept both low- and high-level guidance. Most work in this space has focused on obtaining point estimates of the target image, with the ultimate goal of approximating literal pixel-wise reconstructions of target images from the brain activity patterns they evoke. This emphasis belies the fact that there is always a family of images that are equally compatible with any evoked brain activity pattern, and the fact that many image-generators are inherently stochastic and do not by themselves offer a method for selecting the single best reconstruction from among the samples they generate. We introduce a novel reconstruction procedure (Second Sight) that iteratively refines an image distribution to explicitly maximize the alignment between the predictions of a voxel-wise encoding model and the brain activity patterns evoked by any target image. We show that our process converges on a distribution of high-quality reconstructions by refining both semantic content and low-level image details across iterations. Images sampled from these converged image distributions are competitive with state-of-the-art reconstruction algorithms. Interestingly, the time-to-convergence varies systematically across visual cortex, with earlier visual areas generally taking longer and converging on narrower image distributions, relative to higher-level brain areas. Second Sight thus offers a succinct and novel method for exploring the diversity of representations across visual brain areas., Comment: 15 Figures, 19 pages including the appendix

Visual reconstruction algorithms are an interpretive tool that map brain activity to pixels. Past reconstruction algorithms employed brute-force search through a massive library to select candidate images that, when passed through an encoding model, accurately predict brain activity. Here, we use conditional generative diffusion models to extend and improve this search-based strategy. We decode a semantic descriptor from human brain activity (7T fMRI) in voxels across most of visual cortex, then use a diffusion model to sample a small library of images conditioned on this descriptor. We pass each sample through an encoding model, select the images that best predict brain activity, and then use these images to seed another library. We show that this process converges on high-quality reconstructions by refining low-level image details while preserving semantic content across iterations. Interestingly, the time-to-convergence differs systematically across visual cortex, suggesting a succinct new way to measure the diversity of representations across visual brain areas., Comment: 4 pages, 5 figures, submitted to the 2023 Conference on Cognitive Computational Neuroscience

In this paper we prove that a large class of linear evolution PDEs defines a Stokes-Dirac structure over Hilbert spaces. To do so, the theory of boundary control system is employed. This definition encompasses problems from mechanics, that cannot be handled by the seminal geometric setting given in [van der Schaft and Maschke, Hamiltonian formulation of distributed-parameter systems with boundary energy flow, 2002 ]. Many worked-out examples stemming from continuum mechanics and physics are presented in detail, and a particular focus is given on the functional spaces in duality at the boundary of the geometrical domain. For each example, the connection between the differential operators and the associated Hilbert complexes is illustrated.

This note proposes a paradigm and coordinate system that extends flat, four-dimensional Minkowski spacetime to a broader framework that identifies an event not only in space and in time, but also in terms of possible world, with a third category of coordinates called ``hap'' modelling contingency and counterfactuals. Semantically, hap is based on a Bohmian-like configuration space, in which initial conditions at a specific moment in time uniquely identify a trajectory. This framework can be used to support reasonings that rigorously distinguish between causal dependencies (through time), statistical dependencies (through space) and counterfactual dependencies (through hap). As an example, we use this framework to show that the assumption of free choice is not absolute, but rather depends on the chosen frame of reference: while Alice may see a choice made freely, which is formally a unilateral, single-coordinate translation in hap, Bob sitting in another reference frame might see this same choice not made freely and observe instead a translation in hap jointly across multiple coordinates, which indicates a counterfactual dependency. The framework is agnostic regarding which flavour of the De-Broglie-Bohm theory is considered, as it only makes general assumptions on the set of trajectories that are commonly accepted in the Bohmian mechanics community. It is also general enough to support reasoning on theories for which the separation between possible worlds might also depend on the frame of reference, i.e., two observers might disagree on whether two spacetimehap events happen in the same world or not., Comment: Technical note, 24 pages

An acoustic black hole (ABH) consists of a tapered structure whose thickness follows a power-law profile. When attached to a host structure, an ABH localizes and traps the vibrational energy, which can then be dissipated through, e.g., a damping layer. However, effective vibration mitigation is known to occur only above a cut-on frequency which is inversely proportional to the length of the tapered structure. In this context, the main thrust of this paper is to replace a mechanical ABH by a digital controller so as to create a so-called virtual acoustic black hole (VABH), thus, freeing the ABH from possible mechanical constraints (e.g., compactness, manufacturing and fatigue issues). The proposed VABH is first detailed theoretically. The salient features and performance of the VABH are then demonstrated both numerically and experimentally using a cantilever beam as a host structure. Eventually, it is shown that the VABH significantly enlarges the applicability of the concept of an ABH.

While the emergence of large administrative claims data provides opportunities for research, their use remains limited by the lack of clinical annotations relevant to disease outcomes, such as recurrence in breast cancer (BC). Several challenges arise from the annotation of such endpoints in administrative claims, including the need to infer both the occurrence and the date of the recurrence, the right-censoring of data, or the importance of time intervals between medical visits. Deep learning approaches have been successfully used to label temporal medical sequences, but no method is currently able to handle simultaneously right-censoring and visit temporality to detect survival events in medical sequences. We propose EDEN (Event DEtection Network), a time-aware Long-Short-Term-Memory network for survival analyses, and its custom loss function. Our method outperforms several state-of-the-art approaches on real-world BC datasets. EDEN constitutes a powerful tool to annotate disease recurrence from administrative claims, thus paving the way for the massive use of such data in BC research., Comment: Extended Abstract presented at Machine Learning for Health (ML4H) symposium 2022, November 28th, 2022, New Orleans, United States & Virtual, http://www.ml4h.cc, 6 pages

Numerical validation is at the core of machine learning research as it allows to assess the actual impact of new methods, and to confirm the agreement between theory and practice. Yet, the rapid development of the field poses several challenges: researchers are confronted with a profusion of methods to compare, limited transparency and consensus on best practices, as well as tedious re-implementation work. As a result, validation is often very partial, which can lead to wrong conclusions that slow down the progress of research. We propose Benchopt, a collaborative framework to automate, reproduce and publish optimization benchmarks in machine learning across programming languages and hardware architectures. Benchopt simplifies benchmarking for the community by providing an off-the-shelf tool for running, sharing and extending experiments. To demonstrate its broad usability, we showcase benchmarks on three standard learning tasks: $\ell_2$-regularized logistic regression, Lasso, and ResNet18 training for image classification. These benchmarks highlight key practical findings that give a more nuanced view of the state-of-the-art for these problems, showing that for practical evaluation, the devil is in the details. We hope that Benchopt will foster collaborative work in the community hence improving the reproducibility of research findings., Comment: Accepted in proceedings of NeurIPS 22; Benchopt library documentation is available at https://benchopt.github.io/

We define a specialization map between cohomology algebras of quiver Grassmannians of Dynkin type and we prove that it is surjective in type A. This generalizes a result of Lanini and Strickland., Comment: The new version include a new definition of the specialization map for quiver Grassmannians with respect to the one given in the first version. The new definition can be applied to a large class of quiver algebras

We characterise the symplectic Weyl group elements such that the FFLV basis is compatible with the PBW filtration on symplectic Demazure modules, extending type A results by the second author. Surprisingly, the number of such elements depends not on the type A or C of the Lie algebra but on the rank only., Comment: 20 pages

Time series (TS) are present in many fields of knowledge, research, and engineering. The processing and analysis of TS are essential in order to extract knowledge from the data and to tackle forecasting or predictive maintenance tasks among others The modeling of TS is a challenging task, requiring high statistical expertise as well as outstanding knowledge about the application of Data Mining(DM) and Machine Learning (ML) methods. The overall work with TS is not limited to the linear application of several techniques, but is composed of an open workflow of methods and tests. These workflow, developed mainly on programming languages, are complicated to execute and run effectively on different systems, including Cloud Computing (CC) environments. The adoption of CC can facilitate the integration and portability of services allowing to adopt solutions towards services Internet Technologies (IT) industrialization. The definition and description of workflow services for TS open up a new set of possibilities regarding the reduction of complexity in the deployment of this type of issues in CC environments. In this sense, we have designed an effective proposal based on semantic modeling (or vocabulary) that provides the full description of workflow for Time Series modeling as a CC service. Our proposal includes a broad spectrum of the most extended operations, accommodating any workflow applied to classification, regression, or clustering problems for Time Series, as well as including evaluation measures, information, tests, or machine learning algorithms among others., Comment: 11 pages, 12 figures

Lakehouse systems have reached in the past few years unprecedented size and heterogeneity and have been embraced by many industry players. However, they are often difficult to use as they lack the declarative language and optimization possibilities of relational engines. This paper introduces RumbleML, a high-level, declarative library integrated into the RumbleDB engine and with the JSONiq language. RumbleML allows using a single platform for data cleaning, data preparation, training, and inference, as well as management of models and results. It does it using a purely declarative language (JSONiq) for all these tasks and without any performance loss over existing platforms (e.g. Spark). The key insights of the design of RumbleML are that training sets, evaluation sets, and test sets can be represented as homogeneous sequences of flat objects; that models can be seamlessly embodied in function items mapping input test sets into prediction-augmented result sets; and that estimators can be seamlessly embodied in function items mapping input training sets to models. We argue that this makes JSONiq a viable and seamless programming language for data lakehouses across all their features, whether database-related or machine-learning-related. While lakehouses bring Machine Learning and Data Wrangling on the same platform, RumbleML also brings them to the same language, JSONiq. In the paper, we present the first prototype and compare its performance to Spark showing the benefit of a huge functionality and productivity gain for cleaning up, normalizing, validating data, feeding it into Machine Learning pipelines, and analyzing the output, all within the same system and language and at scale., Comment: 8 pages + references

In this paper, we combine two independent detection methods for identifying fake news: the algorithm VAGO uses semantic rules combined with NLP techniques to measure vagueness and subjectivity in texts, while the classifier FAKE-CLF relies on Convolutional Neural Network classification and supervised deep learning to classify texts as biased or legitimate. We compare the results of the two methods on four corpora. We find a positive correlation between the vagueness and subjectivity measures obtained by VAGO, and the classification of text as biased by FAKE-CLF. The comparison yields mutual benefits: VAGO helps explain the results of FAKE-CLF. Conversely FAKE-CLF helps us corroborate and expand VAGO's database. The use of two complementary techniques (rule-based vs data-driven) proves a fruitful approach for the challenging problem of identifying fake news., Comment: Paper to appear in the Proceedings of the 24th International Conference on Information Fusion. Johannesburg. (2nd version: Typo corrected in metadata in one of the authors' names)

In this letter, we point to three widely accepted challenges that the quantum theory, quantum information, and quantum foundations communities are currently facing: indeterminism, the semantics of conditional probabilities, and the spooky action at a distance. We argue that these issues are fundamentally rooted in conflations commonly made between causal dependencies, counterfactual dependencies, and statistical dependencies. We argue that a simple, albeit somewhat uncomfortable shift of viewpoint leads to a way out of the impossibility to extend the theory beyond indeterminism, and towards the possibility that sound extensions of quantum theory, possibly even deterministic yet not super-deterministic, will emerge in the future. The paradigm shift, which we present here, involves a non-trivial relaxation of the commonly accepted mathematical definition of free choice, leading to non-Nashian free choice, more care with the choice of probabilistic notations, and more rigorous use of vocabulary related to causality, counterfactuals, and correlations, which are three concepts of a fundamentally different nature., Comment: Note. 18 pages

Frequent violations of fair principles in real-life settings raise the fundamental question of whether such principles can guarantee the existence of a self-enforcing equilibrium in a free economy. We show that elementary principles of distributive justice guarantee that a pure-strategy Nash equilibrium exists in a finite economy where agents freely (and non-cooperatively) choose their inputs and derive utility from their pay. Chief among these principles is that: 1) your pay should not depend on your name, and 2) a more productive agent should not earn less. When these principles are violated, an equilibrium may not exist. Moreover, we uncover an intuitive condition -- technological monotonicity -- that guarantees equilibrium uniqueness and efficiency. We generalize our findings to economies with social justice and inclusion, implemented in the form of progressive taxation and redistribution, and guaranteeing a basic income to unproductive agents. Our analysis uncovers a new class of strategic form games by incorporating normative principles into non-cooperative game theory. Our results rely on no particular assumptions, and our setup is entirely non-parametric. Illustrations of the theory include applications to exchange economies, surplus distribution in a firm, contagion and self-enforcing lockdown in a networked economy, and bias in the academic peer-review system. Keywords: Market justice; Social justice; Inclusion; Ethics; Discrimination; Self-enforcing contracts; Fairness in non-cooperative games; Pure strategy Nash equilibrium; Efficiency. JEL Codes: C72, D30, D63, J71, J38

We present Clinica (www.clinica.run), an open-source software platform designed to make clinical neuroscience studies easier and more reproducible. Clinica aims for researchers to i) spend less time on data management and processing, ii) perform reproducible evaluations of their methods, and iii) easily share data and results within their institution and with external collaborators. The core of Clinica is a set of automatic pipelines for processing and analysis of multimodal neuroimaging data (currently, T1-weighted MRI, diffusion MRI and PET data), as well as tools for statistics, machine learning and deep learning. It relies on the brain imaging data structure (BIDS) for the organization of raw neuroimaging datasets and on established tools written by the community to build its pipelines. It also provides converters of public neuroimaging datasets to BIDS (currently ADNI, AIBL, OASIS and NIFD). Processed data include image-valued scalar fields (e.g. tissue probability maps), meshes, surface-based scalar fields (e.g. cortical thickness maps) or scalar outputs (e.g. regional averages). These data follow the ClinicA Processed Structure (CAPS) format which shares the same philosophy as BIDS. Consistent organization of raw and processed neuroimaging files facilitates the execution of single pipelines and of sequences of pipelines, as well as the integration of processed data into statistics or machine learning frameworks. The target audience of Clinica is neuroscientists or clinicians conducting clinical neuroscience studies involving multimodal imaging, and researchers developing advanced machine learning algorithms applied to neuroimaging data.

The stability of a piezoelectric structure controlled by a digital vibration absorber emulating a shunt circuit is investigated in this work. The formalism of feedback control theory is used to demonstrate that systems with a low electromechanical coupling are prone to delay-induced instabilities entailed by the sampling procedure of the digital unit. An explicit relation is derived between the effective electromechanical coupling factor and the maximum sampling period guaranteeing a stable controlled system. Since this sampling period may be impractically small, a simple modification procedure of the emulated admittance of the shunt circuit is proposed in order to counteract the effect of delays by anticipation. The theoretical developments are experimentally validated on a clamped-free piezoelectric beam.

In the domain of high-energy physics (HEP), query languages in general and SQL in particular have found limited acceptance. This is surprising since HEP data analysis matches the SQL model well: the data is fully structured and queried using mostly standard operators. To gain insights on why this is the case, we perform a comprehensive analysis of six diverse, general-purpose data processing platforms using an HEP benchmark. The result of the evaluation is an interesting and rather complex picture of existing solutions: Their query languages vary greatly in how natural and concise HEP query patterns can be expressed. Furthermore, most of them are also between one and two orders of magnitude slower than the domain-specific system used by particle physicists today. These observations suggest that, while database systems and their query languages are in principle viable tools for HEP, significant work remains to make them relevant to HEP researchers., Comment: This is the extended version of a full paper to appear in PVLDB 15.2 (VLDB 2022)

Structural optimization (topology, shapes, sizing) is an important tool for facilitating the emergence of new concepts in structural design. Normally, topology optimization is carried out at the early stage of design and then shape and sizing design are performed sequentially. Unlike traditional topology optimization method, explicit methodologies have attracted a great deal of attention because of the advantages of shortcuting the costly CAD/CAE processes while dealing with low order number of design variables compared to implicit method (such as SIMP). This paper aims at presenting an adaptation of a flow-inspired approach so-called Moving Node Approach (MNA) in topology optimization. In this approach, the discretization is decoupled from the material distribution and the final objective is to recognize the best beam assembly while minimizing compliance. The paradigm has here changed and new design variables are used such as nodes location, elements length/orientation and size providing a lower number of design variables than pixels-based. The methodology is validated using 2 classical testcases in the field of Topology Optimization: the Cantilever beam and the L-Shape problem.