All Years Seminars
[INGI] 2026-03-05 (13:00) : xPUBench: Scalable and Energy-Efficient GPU and DPU-Accelerated Network Functions
At Nyquist Maxwell a.164
Speaker :
Maxime Vanliefde (ICTEAM)
Abstract : The rapid increase in network speeds makes packet processing on general-purpose CPUs increasingly challenging. At 100 Gbps and beyond, CPUs struggle to sustain complex network functions without dedicated acceleration. This trend motivates the exploration and measurement of alternative compute platforms such as GPUs and embedded CPUs in Network Interface Cards (NICs). Modern NICs provide tighter integration with GPUs, with the ability to write received packets directly to GPU memory. SmartNICs, also known as DPUs, further feature embedded ARM or RISC cores capable of offloading NFV packet processing entirely. In this work, we introduce xPUBench, a benchmarking environment that systematically measures the performance and energy efficiency of packet processing across CPUs, GPUs, and DPUs. We evaluate several (co-)processing models relevant to Network Function Virtualization, including CPU+GPU hybrid, DPU-only, and GPU-only approaches. Our measurements show that, for a computation-heavy workload, current CPU-only implementations manage to handle up to 50% of the 100 Gbps NIC rate. In contrast, GPU implementations can saturate it. We also show that SmartNICs’ most powerful embedded cores can replace the main CPU for some traditional packet processing, alleviating the load on the host, which can now be entirely dedicated to running applications. We finally propose a novel energy-efficiency dimension, showing that DPUs outperform traditional CPUs for low-throughput processing, requiring only 24 W to sustain 10 Gbps, and that GPUs outperform CPUs for high-throughput processing. Our findings emphasize the need to assess both performance and energy in heterogeneous packet-processing pipelines, given the growing diversity of “xPUs” in networked systems.
[INMA] 2026-03-03 (14:00) : Consensus is a myth: Human label variation in Natural Language Inference
At Euler building (room A.002)
Speaker :
Marie-Catherine de Marneffe (CENTAL, UCLouvain)
Abstract : Recently, NLP researchers have increasingly begun to acknowledge that humans often diverge in their interpretations of various NLP tasks, and that such variation should be captured if robust language understanding is to be achieved. In this talk, I will focus on analyzing human label variation in the Natural Language Inference (NLI) task, in which, given a premise, one identifies whether a hypothesis sentence is true, false, or undetermined. For instance, if one says: “My friend often travels with a heavy suitcase”, can it be inferred that “My friend often travels with a light suitcase”? I will examine the various sources of NLI label variation and investigate whether or not they can be captured by current LLMs, arguing that, in the presence of variation, labels without explanations are not sufficiently meaningful.
[LINGI2399] 2026-02-26 (10:45) : Applying Generative AI and NLP to Historical Texts
At BARB94
Speaker :
Xavier Gillard (UCLouvain)
Abstract : The presentation introduces the Arkey project, a long-term applied research collaboration between UCLouvain and the Belgian State Archives. It aims at improving how users (archivists, researchers and the public) interact with archival collections through computational methods. A central component is a comparative study of Named Entity Recognition on noisy XVth-XVIIth century texts, contrasting general-purpose Large Language Models with specialiwed, fine-tuned encoders such as XLM-RoBERTa. Using a mixed evaluation framework that combines standard NLP metrics with a human preference study, the project shows that expert users prioritize the factual accuracy of smaller specialized models more than the fluent but sometimes hallucinatory outputs of LLMs.
The talk also presents “Ask Agatha,” an agentic retrieval-augmented generation 5RAG) system developed for the national archives. We demonstrate the necessity for moving beyond simple RAG pipelines toward a stateful, graph-based agent architecture capable of complex, multi-step tool interactions. We detail experiments such as cascade patterns for efficient tool use and the development of “digita,” a fine-tuned 8B model designed to emulate the archivists’ expert communication style. The presentation concludes by synthesizing lessons learned on model specialization, agentic architecture, and evaluation strategies, illustrating the solutions developed to meet the needs of a specialized expert domain.
[INGI] 2026-02-25 (11:00) : Closed Loop Automation System for xG networks
At Shannon
Speaker :
Stefano Secci (CNAM, France)
,
and
Patient NTUMBA WA NTUMBA (CNAM Paris)
Abstract : In this presentation, we present a closed-loop automation system that leverages in-network distributed learning to automatically mitigate anomalous states in the connect-compute software infrastructure. We describe the key components of the automation system, including: an anomaly detection module based on federated learning; AI function scheduling to meet detection performance targets while mitigating federated learning stragglers; a data-pipeline design that supports high accuracy, real-time preprocessing, and timely data delivery; a network data sources load-balancing strategy for federated-learning clients; and automated reconfiguration management using deep reinforcement learning (DRL). We also deliver a live demo of the core system blocks—showing, in real time, the anomalies detected during the inference phase using federated learning with real-time data pipelining.
[ICTM] 2026-02-24 (13:00) : Distributed, Coordination-Free Programming: 10 Years of Progress Since Lasp
At Shannon Room, Maxwell Building, 3 Place du Levant, Louvain-la-Neuve
Speaker :
Peter VAN ROY (UCL)
Abstract : Consensus is a critical building block for building fault-tolerant distributed systems. It is widely believed that without consensus, large distributed applications on the Internet could not exist. But recent advances show that consistent replication can be achieved without consensus by using convergent data structures such as CRDTs (conflict-free replicated data types). This is called coordination-free programming and it has become a credible alternative to consensus. The Lasp system is the first to compose CRDTs. It was published in 2015 in the ACM Symposium on Principles and Practice of Declarative Programming (PPDP) and the paper won the 10-year most influential paper award at PPDP 2025. Lasp’s coordination-free model has inspired a decade of progress in academia and industry. As the industry shifts toward multi-region deployments, Lasp’s core insight — that coordination can be the exception, not the rule — continues to shape how we build reliable, scalable systems today.
Seminars
Subscribe (ICS)
INMA Contact Info
Mathematical Engineering (INMA)
L4.05.01
Avenue Georges Lemaître, 4
+32 10 47 80 36
secretaire-inma@uclouvain.be
Mon – Fri 9:00A.M. – 5:00P.M.
JOIN US