arXiv:2512.05567v1 Announce Type: cross 
Abstract: The main objective of this study is to propose an optimal transport based semi-supervised approach to learn from scarce labelled image data using deep convolutional networks. The principle lies in implicit graph-based transductive semi-supervised learning where the similarity metric between image samples is the Wasserstein distance. This metric is used in the label propagation mechanism during learning. We apply and demonstrate the effectiveness of the method on a GNSS real life application. More specifically, we address the problem of multi-path interference detection. Experiments are conducted under various signal conditions. The results show that for specific choices of hyperparameters controlling the amount of semi-supervision and the level of sensitivity to the metric, the classification accuracy can be significantly improved over the fully supervised training method.

قدمت دراسة جديدة نهجًا للتعلم شبه المراقب يستخدم مسافة فاسرشتاين لتحسين دقة التصنيف في الشبكات العصبية التلافيفية العميقة، خاصةً للكشف عن تداخل المسارات المتعددة في تطبيقات GNSS. تعتمد الطريقة على إطار عمل للتعلم الانتقائي القائم على الرسوم البيانية لنشر العلامات بفعالية من بيانات محدودة.

Un nuevo estudio ha presentado un enfoque de aprendizaje semi-supervisado que utiliza la distancia de Wasserstein para mejorar la precisión de clasificación en redes neuronales convolucionales profundas, especialmente para la detección de interferencias de múltiples trayectorias en aplicaciones GNSS. El método aprovecha un marco de aprendizaje transductivo basado en grafos para propagar etiquetas de manera efectiva a partir de datos etiquetados limitados.

Une nouvelle étude a introduit une approche d'apprentissage semi-supervisé utilisant la distance de Wasserstein pour améliorer la précision de classification dans les réseaux de neurones convolutionnels profonds, en particulier pour la détection des interférences multi-trajets dans les applications GNSS. La méthode s'appuie sur un cadre d'apprentissage transductif basé sur des graphes pour propager efficacement les étiquettes à partir de données étiquetées limitées.

A new study has introduced a semi-supervised learning approach utilizing the Wasserstein distance to enhance classification accuracy in deep convolutional networks, particularly for detecting multi-path interference in GNSS applications. The method leverages a graph-based transductive learning framework to propagate labels effectively from limited labeled data.

Wasserstein distance based semi-supervised manifold learning and application to GNSS multi-path detection

arXiv:2512.09530v1 Announce Type: new 
Abstract: This thesis examines self-attention training through the lens of Optimal Transport (OT) and develops an OT-based alternative for tabular classification. The study tracks intermediate projections of the self-attention layer during training and evaluates their evolution using discrete OT metrics, including Wasserstein distance, Monge gap, optimality, and efficiency. Experiments are conducted on classification tasks with two and three classes, as well as on a biomedical dataset.
  Results indicate that the final self-attention mapping often approximates the OT optimal coupling, yet the training trajectory remains inefficient. Pretraining the MLP section on synthetic data partially improves convergence but is sensitive to their initialization. To address these limitations, an OT-based algorithm is introduced: it generates class-specific dummy Gaussian distributions, computes an OT alignment with the data, and trains an MLP to generalize this mapping. The method achieves accuracy comparable to Transformers while reducing computational cost and scaling more efficiently under standardized inputs, though its performance depends on careful dummy-geometry design. All experiments and implementations are conducted in R.

تدرس أطروحة حديثة تدريب الانتباه الذاتي لتصنيف البيانات الجدولية من خلال النقل الأمثل (OT)، حيث تطور بديلاً يعتمد على OT يتتبع تطور طبقات الانتباه الذاتي أثناء التدريب باستخدام مقاييس OT المنفصلة مثل مسافة فاسرشتاين وفجوة مونج. تكشف الدراسة أنه على الرغم من أن خريطة الانتباه الذاتي النهائية تقترب من الاقتران الأمثل لـ OT، إلا أن عملية التدريب تظل غير فعالة.

Una reciente tesis explora el entrenamiento de la atención automática para la clasificación tabular a través del Transporte Óptimo (OT), desarrollando una alternativa basada en OT que rastrea la evolución de las capas de atención automática durante el entrenamiento utilizando métricas discretas de OT como la distancia de Wasserstein y la brecha de Monge. El estudio revela que, aunque la mapeo final de la atención automática se aproxima al acoplamiento óptimo de OT, el proceso de entrenamiento sigue siendo ineficiente.

Une récente thèse examine l'entraînement de l'attention automatique pour la classification tabulaire à travers le Transport Optimal (OT), développant une alternative basée sur l'OT qui suit l'évolution des couches d'attention automatique pendant l'entraînement à l'aide de métriques OT discrètes telles que la distance de Wasserstein et l'écart de Monge. L'étude révèle que, bien que la cartographie finale de l'attention automatique approche le couplage optimal de l'OT, le processus d'entraînement reste inefficace.

A recent thesis explores self-attention training for tabular classification through Optimal Transport (OT), developing an OT-based alternative that tracks the evolution of self-attention layers during training using discrete OT metrics like Wasserstein distance and Monge gap. The study reveals that while the final self-attention mapping approximates the OT optimal coupling, the training process remains inefficient.

Wasserstein distance based semi-supervised manifold learning and application to GNSS multi-path detection

Was this article worth reading? Share it

LucidQuery AI

WasItAI

Attentive AI

Ready to build your own newsroom?