Searcharxiv⌕ Search

arXiv subjects

Julien Lagarde

Publications and source records attributed to Julien Lagarde.

8 recordsLinked to original sources
arXiv2025

The hands of time: Moving my body to keep time order in the brain

The brain is very often viewed as a network, be it at small scale made of cells, mostly neurons, or at larger scale made of neuronal assemblies. Here we introduce a conjecture, in the spirit of a philosophical though experiment, which proposes that the present cannot be obtained from within such networks, and that this limitation imposes burdens on network efficiency in information processing. We aim to argue this conjecture imposes recurrent contacts from within the brain to outside in the physical world via behavior, which create a flow of time stamps. This though experiment may contribute to make the divide between the foci toward inside versus outside, for example opposing ecological psychology and many frameworks adopted in neurosciences, superfluous. This piece proposes an ambulation triggered by a thought experiment: What if I was a neuron listening to another one and talking to a third? It is a modest attempt to walk in the footsteps of classical thought experiments, like Molyneux problem, the imitation game and the anti-sequel Chinese room, key gedankenexperiments in an elevator in physics, or the cogito in philosophy.

q-bio.NC↗
arXiv2024

Two-step registration method boosts sensitivity in longitudinal fixel-based analyses

Longitudinal analyses are increasingly used in clinical studies as they allow the study of subtle changes over time within the same subjects. In most of these studies, it is necessary to align all the images studied to a common reference by registering them to a template. In the study of white matter using the recently developed fixel-based analysis (FBA) method, this registration is important, in particular because the fiber bundle cross-section metric is a direct measure of this registration. In the vast majority of longitudinal FBA studies described in the literature, sessions acquired for a same subject are directly independently registered to the template. However, it has been shown in T1-based morphometry that a 2-step registration through an intra-subject average can be advantageous in longitudinal analyses. In this work, we propose an implementation of this 2-step registration method in a typical longitudinal FBA aimed at investigating the evolution of white matter changes in Alzheimer's disease (AD). We compared at the fixel level the mean absolute effect and standard deviation yielded by this registration method and by a direct registration, as well as the results obtained with each registration method for the study of AD in both fixelwise and tract-based analyses. We found that the 2-step method reduced the variability of the measurements and thus enhanced statistical power in both types of analyses.

q-bio.QM↗
arXiv2024

Transformer with Controlled Attention for Synchronous Motion Captioning

In this paper, we address a challenging task, synchronous motion captioning, that aim to generate a language description synchronized with human motion sequences. This task pertains to numerous applications, such as aligned sign language transcription, unsupervised action segmentation and temporal grounding. Our method introduces mechanisms to control self- and cross-attention distributions of the Transformer, allowing interpretability and time-aligned text generation. We achieve this through masking strategies and structuring losses that push the model to maximize attention only on the most important frames contributing to the generation of a motion word. These constraints aim to prevent undesired mixing of information in attention maps and to provide a monotonic attention distribution across tokens. Thus, the cross attentions of tokens are used for progressive text generation in synchronization with human motion sequences. We demonstrate the superior performance of our approach through evaluation on the two available benchmark datasets, KIT-ML and HumanML3D. As visual evaluation is essential for this task, we provide a comprehensive set of animated visual illustrations in the code repository: https://github.com/rd20karim/Synch-Transformer.

cs.CV↗
arXiv2024

Guided Attention for Interpretable Motion Captioning

Diverse and extensive work has recently been conducted on text-conditioned human motion generation. However, progress in the reverse direction, motion captioning, has seen less comparable advancement. In this paper, we introduce a novel architecture design that enhances text generation quality by emphasizing interpretability through spatio-temporal and adaptive attention mechanisms. To encourage human-like reasoning, we propose methods for guiding attention during training, emphasizing relevant skeleton areas over time and distinguishing motion-related words. We discuss and quantify our model's interpretability using relevant histograms and density distributions. Furthermore, we leverage interpretability to derive fine-grained information about human motion, including action localization, body part identification, and the distinction of motion-related words. Finally, we discuss the transferability of our approaches to other tasks. Our experiments demonstrate that attention guidance leads to interpretable captioning while enhancing performance compared to higher parameter-count, non-interpretable state-of-the-art systems. The code is available at: https://github.com/rd20karim/M2T-Interpretable.

cs.CV↗
arXiv2023

Motion2Language, unsupervised learning of synchronized semantic motion segmentation

In this paper, we investigate building a sequence to sequence architecture for motion to language translation and synchronization. The aim is to translate motion capture inputs into English natural-language descriptions, such that the descriptions are generated synchronously with the actions performed, enabling semantic segmentation as a byproduct, but without requiring synchronized training data. We propose a new recurrent formulation of local attention that is suited for synchronous/live text generation, as well as an improved motion encoder architecture better suited to smaller data and for synchronous generation. We evaluate both contributions in individual experiments, using the standard BLEU4 metric, as well as a simple semantic equivalence measure, on the KIT motion language dataset. In a follow-up experiment, we assess the quality of the synchronization of generated text in our proposed approaches through multiple evaluation metrics. We find that both contributions to the attention mechanism and the encoder architecture additively improve the quality of generated text (BLEU and semantic equivalence), but also of synchronization. Our code is available at https://github.com/rd20karim/M2T-Segmentation/tree/main

cs.CV↗
arXiv2019

When is now in a distributed system? Animated motion (could) set the present in brain networks

Our brains are viewed as interconnected distributed systems. The connections between distant areas in the brain are significantly delayed. How to obtained now in such networks with delayed interconnections? We aim to show that delayed communication and interconnectedness of the brain impose an interaction with the environment, assuming that such an access to now, which we label t-present, is of use for this system. It is conjectured that for any sensory, motor or cognitive functions to work efficiently an updated sort of time origin is required, and we claim that it is uniquely given by a direct contact with the physical environment. To get such contact autonomously any movement is required, be it originating in the motion of sensory systems or in goal directed movements. Some limit cases are identified and discussed. Next, several testable situations are envisioned and available studies in favor of the main theoretical hypothesis are shortly reviewed. Finally, as a proof of concept, an experimental study employing galvanic vestibular stimulation is presented and discussed.

q-bio.NC↗
arXiv2017

To do things with words (only): An introduction to the role of noise in coordination dynamics without equations

Uncertainty, spatial or temporal errors, variability, are classic themes in the study of human and animal behaviors. Several theoretical approaches1 and concepts have been adopted to tackle those issues, often considering the CNS as an observer, using Shannon information and entropy, signal to noise ratio, and recently a Bayesian approach, and free energy minimization. In the coordination dynamics framework, addressing pattern formation processes underlying cognitive functions, and goal directed human movement among others, the tools employed originate from the statistical physics of Brownian motion and stochastic processes. The relations between those theories/ concepts have been drawn for some cases in their original fields, for example between free energy minimization and diffusion model in a force field (Jordan et al., 1998). Here an introductory presentation of the approach of noise in coordination dynamics is proposed, aimed at the experimentalist.

q-bio.NC↗