arXiv subjects

Hugo Thimonier

Publications and source records attributed to Hugo Thimonier.

10 recordsLinked to original sources

arXiv2026

Robostral Navigate

Deploying navigation systems at scale requires a recipe that minimizes sensor assumptions, generalizes across robot embodiments, and trains efficiently. Yet, today's best systems depend on depth sensors, multi-camera rigs, or pre-built maps, limiting the hardware they support and increasing deployment cost. We introduce Robostral Navigate, an 8B vision-language model built around this scalability objective. The model consumes only a stream of monocular RGB images - the most ubiquitous sensor across robotic platforms and predicts waypoints by pointing to the next target location in the current camera view. Operating purely in image space, rather than robot-specific coordinates, makes the policy naturally robust to changes in camera intrinsics and scene scale, enabling deployment across wheeled, legged, and aerial robots without recalibration. We generate 2.4 million trajectories across 350k simulated scenes to reduce the reliance on real-world data collection and scale easily. We further introduce a prefix-caching training recipe that packs entire episodes into single training sequences, reducing training tokens by 22x and cutting training time from months to days. A tree-based attention mask prevents conditioning on previous ground-truth actions, encouraging visually grounded action prediction, and reinforcement learning is used to further improve exploration and recovery capabilities. On the Room-to-Room and Room-Across-Room in Continuous Environments (R2R-CE and RxR-CE) benchmarks, Robostral Navigate sets a new state of the art. On R2R-CE, it achieves a 77.4% success rate, surpassing the best monocular method by 10.5 points and the strongest depth- or multi-camera system by 5.3 points despite using only a single RGB camera. On RxR-CE, it reaches 75.1% success rate, outperforming all monocular baselines.

cs.RO↗

arXiv2026

Voxtral TTS

We introduce Voxtral TTS, an expressive multilingual text-to-speech model that generates natural speech from as little as 3 seconds of reference audio. Voxtral TTS adopts a hybrid architecture that combines auto-regressive generation of semantic speech tokens with flow-matching for acoustic tokens. These tokens are encoded and decoded with Voxtral Codec, a speech tokenizer trained from scratch with a hybrid VQ-FSQ quantization scheme. In human evaluations conducted by native speakers, Voxtral TTS is preferred for multilingual voice cloning due to its naturalness and expressivity, achieving a 68.4\% win rate over ElevenLabs Flash v2.5. We release the model weights under a CC BY-NC license.

cs.AI↗

arXiv2026

Voxtral Realtime

We introduce Voxtral Realtime, a natively streaming automatic speech recognition model that matches offline transcription quality at sub-second latency. Unlike approaches that adapt offline models through chunking or sliding windows, Voxtral Realtime is trained end-to-end for streaming, with explicit alignment between audio and text streams. Our architecture builds on the Delayed Streams Modeling framework, introducing a new causal audio encoder and Ada RMS-Norm for improved delay conditioning. We scale pretraining to a large-scale dataset spanning 13 languages. At a delay of 480ms, Voxtral Realtime achieves performance on par with Whisper, the most widely deployed offline transcription system. We release the model weights under the Apache 2.0 license.

cs.AI↗

arXiv2025

EmoSLLM: Parameter-Efficient Adaptation of LLMs for Speech Emotion Recognition

Emotion recognition from speech is a challenging task that requires capturing both linguistic and paralinguistic cues, with critical applications in human-computer interaction and mental health monitoring. Recent works have highlighted the ability of Large Language Models (LLMs) to perform tasks outside of the sole natural language area. In particular, recent approaches have investigated coupling LLMs with other data modalities by using pre-trained backbones and different fusion mechanisms. This work proposes a novel approach that fine-tunes an LLM with audio and text representations for emotion prediction. Our method first extracts audio features using an audio feature extractor, which are then mapped into the LLM's representation space via a learnable interfacing module. The LLM takes as input (1) the transformed audio features, (2) additional features in the form of natural language (e.g., the transcript), and (3) a textual prompt describing the emotion prediction task. To efficiently adapt the LLM to this multimodal task, we employ Low-Rank Adaptation (LoRA), enabling parameter-efficient fine-tuning. Experimental results on standard emotion recognition benchmarks demonstrate that our model outperforms all but one existing Speech-Text LLMs in the literature, while requiring less than half the parameters of competing approaches. This highlights our approach's effectiveness in integrating multi-modal inputs for speech-based emotion understanding while maintaining significant computational efficiency.

eess.AS↗

arXiv2024

T-JEPA: Augmentation-Free Self-Supervised Learning for Tabular Data

Self-supervision is often used for pre-training to foster performance on a downstream task by constructing meaningful representations of samples. Self-supervised learning (SSL) generally involves generating different views of the same sample and thus requires data augmentations that are challenging to construct for tabular data. This constitutes one of the main challenges of self-supervision for structured data. In the present work, we propose a novel augmentation-free SSL method for tabular data. Our approach, T-JEPA, relies on a Joint Embedding Predictive Architecture (JEPA) and is akin to mask reconstruction in the latent space. It involves predicting the latent representation of one subset of features from the latent representation of a different subset within the same sample, thereby learning rich representations without augmentations. We use our method as a pre-training technique and train several deep classifiers on the obtained representation. Our experimental results demonstrate a substantial improvement in both classification and regression tasks, outperforming models trained directly on samples in their original data space. Moreover, T-JEPA enables some methods to consistently outperform or match the performance of traditional methods likes Gradient Boosted Decision Trees. To understand why, we extensively characterize the obtained representations and show that T-JEPA effectively identifies relevant features for downstream tasks without access to the labels. Additionally, we introduce regularization tokens, a novel regularization method critical for training of JEPA-based models on structured data.

cs.LG↗

arXiv2024

Retrieval Augmented Deep Anomaly Detection for Tabular Data

Deep learning for tabular data has garnered increasing attention in recent years, yet employing deep models for structured data remains challenging. While these models excel with unstructured data, their efficacy with structured data has been limited. Recent research has introduced retrieval-augmented models to address this gap, demonstrating promising results in supervised tasks such as classification and regression. In this work, we investigate using retrieval-augmented models for anomaly detection on tabular data. We propose a reconstruction-based approach in which a transformer model learns to reconstruct masked features of \textit{normal} samples. We test the effectiveness of KNN-based and attention-based modules to select relevant samples to help in the reconstruction process of the target sample. Our experiments on a benchmark of 31 tabular datasets reveal that augmenting this reconstruction-based anomaly detection (AD) method with sample-sample dependencies via retrieval modules significantly boosts performance. The present work supports the idea that retrieval module are useful to augment any deep AD method to enhance anomaly detection on tabular data.

cs.LG↗

arXiv2023

Comparative Evaluation of Anomaly Detection Methods for Fraud Detection in Online Credit Card Payments

This study explores the application of anomaly detection (AD) methods in imbalanced learning tasks, focusing on fraud detection using real online credit card payment data. We assess the performance of several recent AD methods and compare their effectiveness against standard supervised learning methods. Offering evidence of distribution shift within our dataset, we analyze its impact on the tested models' performances. Our findings reveal that LightGBM exhibits significantly superior performance across all evaluated metrics but suffers more from distribution shifts than AD methods. Furthermore, our investigation reveals that LightGBM also captures the majority of frauds detected by AD methods. This observation challenges the potential benefits of ensemble methods to combine supervised, and AD approaches to enhance performance. In summary, this research provides practical insights into the utility of these techniques in real-world scenarios, showing LightGBM's superiority in fraud detection while highlighting challenges related to distribution shifts.

cs.LG↗

arXiv2023

Beyond Individual Input for Deep Anomaly Detection on Tabular Data

Anomaly detection is vital in many domains, such as finance, healthcare, and cybersecurity. In this paper, we propose a novel deep anomaly detection method for tabular data that leverages Non-Parametric Transformers (NPTs), a model initially proposed for supervised tasks, to capture both feature-feature and sample-sample dependencies. In a reconstruction-based framework, we train an NPT to reconstruct masked features of normal samples. In a non-parametric fashion, we leverage the whole training set during inference and use the model's ability to reconstruct the masked features to generate an anomaly score. To the best of our knowledge, this is the first work to successfully combine feature-feature and sample-sample dependencies for anomaly detection on tabular datasets. Through extensive experiments on 31 benchmark tabular datasets, we demonstrate that our method achieves state-of-the-art performance, outperforming existing methods by 2.4% and 1.2% in terms of F1-score and AUROC, respectively. Our ablation study further proves that modeling both types of dependencies is crucial for anomaly detection on tabular data.

cs.LG↗

arXiv2022

TracInAD: Measuring Influence for Anomaly Detection

As with many other tasks, neural networks prove very effective for anomaly detection purposes. However, very few deep-learning models are suited for detecting anomalies on tabular datasets. This paper proposes a novel methodology to flag anomalies based on TracIn, an influence measure initially introduced for explicability purposes. The proposed methods can serve to augment any unsupervised deep anomaly detection method. We test our approach using Variational Autoencoders and show that the average influence of a subsample of training points on a test point can serve as a proxy for abnormality. Our model proves to be competitive in comparison with state-of-the-art approaches: it achieves comparable or better performance in terms of detection accuracy on medical and cyber-security tabular benchmark data.

cs.LG↗

arXiv2021

Learning Long-Term Style-Preserving Blind Video Temporal Consistency

When trying to independently apply image-trained algorithms to successive frames in videos, noxious flickering tends to appear. State-of-the-art post-processing techniques that aim at fostering temporal consistency, generate other temporal artifacts and visually alter the style of videos. We propose a postprocessing model, agnostic to the transformation applied to videos (e.g. style transfer, image manipulation using GANs, etc.), in the form of a recurrent neural network. Our model is trained using a Ping Pong procedure and its corresponding loss, recently introduced for GAN video generation, as well as a novel style preserving perceptual loss. The former improves long-term temporal consistency learning, while the latter fosters style preservation. We evaluate our model on the DAVIS and videvo.net datasets and show that our approach offers state-of-the-art results concerning flicker removal, and better keeps the overall style of the videos than previous approaches.

cs.CV↗