Home
About
People
Publications
Seminars
- Seminar Calendar
- Seminar Videos
Workshops
HLT Masters
Resources

Forms

JHU Fly America Act Waiver Checklist

Subscribe to the CLSP Seminars Mailing List

Upcoming Seminars

There are no upcoming events.

Add

Add to Timely Calendar
Add to Google
Add to Outlook
Add to Apple Calendar
Add to other calendar
Export to XML

Seminar Videos

** Due to technical problems, the meeting moved to Zoom, we apologize for the inconvenience **

📍 Live from FIT, Brno University of Technology (Czech Republic), room E112
🕘 July 15th, 2025 — 11:00 CEST
🎙️ Loïc Barrault [Meta AI, France]

Current methods to reach Advanced Machine Intelligence are almost all based on the token-level Large Language Model paradigm. It has attracted a big part of the research community and huge progress have been made. However, we argue that token-based LLMs lack crucial characteristics of human intelligence that limit their potential, such as explicit reasoning and planning, hierarchical processing as well as multilingual processing. In this talk, I will present the Large Concept Model, a model trained to reason over a multimodal and multilingual sentence representation space. This diffusion-based model shows strong performance on several generative tasks and exhibits strong 0-shot multilingual capabilities. Several variants have been explored, including an initial attempt to hierarchical processing of text.

Bio: Loïc Barrault (M) is a Research Scientist at Meta AI. Previously, he was a Senior Lecturer in the NLP group of the University of Sheffield and an Associate Professor at LIUM, University of Le Mans. He obtained his PhD at the University of Avignon in 2008 in the field of automatic speech recognition. His research work focuses on statistical and neural machine translation, by considering multiple modalities (multimodal neural machine translation) and by designing lifelong learning methods for MT. Recent work include text-to-text machine translation for 200 languages (NLLB200), speech-to-speech translation for 100 languages (Seamless-M4T) along its expressive version (Seamless Expressive) and reasoning in the embedding space (Large Concept Model).

JSALT 2025 - Plenary Talk - L.Barrault - Large Concept Model: beyond token-based LLMs

** Due to technical problems, the meeting moved to Zoom! Connect Here: https://cesnet.zoom.us/j/94996152322 **

📍 Live from FIT, Brno University of Technology (Czech Republic), room E112
🕘 July 15th, 2025 — 11:00 CEST
🎙️ Loïc Barrault [Meta AI, France]

Current methods to reach Advanced Machine Intelligence are almost all based on the token-level Large Language Model paradigm. It has attracted a big part of the research community and huge progress have been made. However, we argue that token-based LLMs lack crucial characteristics of human intelligence that limit their potential, such as explicit reasoning and planning, hierarchical processing as well as multilingual processing. In this talk, I will present the Large Concept Model, a model trained to reason over a multimodal and multilingual sentence representation space. This diffusion-based model shows strong performance on several generative tasks and exhibits strong 0-shot multilingual capabilities. Several variants have been explored, including an initial attempt to hierarchical processing of text.

Bio: Loïc Barrault (M) is a Research Scientist at Meta AI. Previously, he was a Senior Lecturer in the NLP group of the University of Sheffield and an Associate Professor at LIUM, University of Le Mans. He obtained his PhD at the University of Avignon in 2008 in the field of automatic speech recognition. His research work focuses on statistical and neural machine translation, by considering multiple modalities (multimodal neural machine translation) and by designing lifelong learning methods for MT. Recent work include text-to-text machine translation for 200 languages (NLLB200), speech-to-speech translation for 100 languages (Seamless-M4T) along its expressive version (Seamless Expressive) and reasoning in the embedding space (Large Concept Model).

JSALT 2025 - Plenary Talk - L.Barrault - Connect Here: https://cesnet.zoom.us/j/94996152322

📍 Live from FIT, Brno University of Technology (Czech Republic), room E112
🕘 July 10th, 2025 — 11:00 CEST
🎙️ Ramani Duraiswami [University of Maryland, College Park]

Learning via deep neural networks has achieved great success in learning functions relating large datasets to their labels in areas like natural language processing, computer vision, and speech processing. It continues to revolutionize many areas of science and society. I will briefly describe various themes of machine learning research underway in my group at UMD: Differentiable Modeling, speeding up the Attention mechanism in Training and Inference in Transformer architectures, and Large Audio Language Models.

The main part of my talk will be focused on the use of differentiable forward modeling in various ways.

An under-appreciated aspect of the deep learning revolution is the use of automatic differentiation and backpropagation on differentiable computational graphs to obtain the parameters specifying the networks. Before learning from data became the method of choice, scientists spent entire careers developing forward models that captured much scientific knowledge about the domains they worked on. These models were based on mathematics, physics, biology, and other scientific areas they worked on. Making these forward models differentiable allows for this knowledge to be incorporated in deep learning architectures. This allows achieving computational pipelines that can incorporate deep learning for tasks like parameter optimization, cost function minimization, inverse problem solution, implicit neural representations, and learning explainable models, that work well in domains where data is sparse.

We apply these ideas in domains like computer graphics, human hearing, room acoustics, signal processing, and in the solution of inverse problems arising in mathematical physics. We will present example solutions and results.

Bio: Ramani Duraiswami is Professor in the department of Computer Science at the University of Maryland, College Park. He also has appointments at Artificial Intelligence Institute at Maryland, UMIACS, Electrical Engineering, Robotics program, Neural and Cognitive Sciences program, and the Applied Math and Scientific Computing Program at the same university. Prof. Duraiswami got his B. Tech. from IIT Bombay and his Ph.D. from The Johns Hopkins University. His research interests are in machine learning, scientific computing and computational perception. Two companies have been spun out based on his research. The audio engine used in content that plays on the millions of shipping VR headsets, PCs, and headphones is based on work from his lab.

[slides] JSALT 2025 - Plenary Talk - Ramani Duraiswami: Differentiable Modeling for Machine Learning

📍 Live from FIT, Brno University of Technology (Czech Republic), room E112
🕘 July 10th, 2025 — 11:00 CEST
🎙️ Ramani Duraiswami [University of Maryland, College Park]

Learning via deep neural networks has achieved great success in learning functions relating large datasets to their labels in areas like natural language processing, computer vision, and speech processing. It continues to revolutionize many areas of science and society. I will briefly describe various themes of machine learning research underway in my group at UMD: Differentiable Modeling, speeding up the Attention mechanism in Training and Inference in Transformer architectures, and Large Audio Language Models.

The main part of my talk will be focused on the use of differentiable forward modeling in various ways.

An under-appreciated aspect of the deep learning revolution is the use of automatic differentiation and backpropagation on differentiable computational graphs to obtain the parameters specifying the networks. Before learning from data became the method of choice, scientists spent entire careers developing forward models that captured much scientific knowledge about the domains they worked on. These models were based on mathematics, physics, biology, and other scientific areas they worked on. Making these forward models differentiable allows for this knowledge to be incorporated in deep learning architectures. This allows achieving computational pipelines that can incorporate deep learning for tasks like parameter optimization, cost function minimization, inverse problem solution, implicit neural representations, and learning explainable models, that work well in domains where data is sparse.

We apply these ideas in domains like computer graphics, human hearing, room acoustics, signal processing, and in the solution of inverse problems arising in mathematical physics. We will present example solutions and results.

Bio: Ramani Duraiswami is Professor in the department of Computer Science at the University of Maryland, College Park. He also has appointments at Artificial Intelligence Institute at Maryland, UMIACS, Electrical Engineering, Robotics program, Neural and Cognitive Sciences program, and the Applied Math and Scientific Computing Program at the same university. Prof. Duraiswami got his B. Tech. from IIT Bombay and his Ph.D. from The Johns Hopkins University. His research interests are in machine learning, scientific computing and computational perception. Two companies have been spun out based on his research. The audio engine used in content that plays on the millions of shipping VR headsets, PCs, and headphones is based on work from his lab.

[camera] JSALT 2025 - Plenary Talk - R.Duraiswami: Differentiable Modeling for Machine Learning

📍 Live from FIT, Brno University of Technology (Czech Republic), room E112
🕘 July 1st, 2025 — 11:00 CEST
🎙️ Barbara Schuppler (TU Graz, Austria)

In recent years, conversational speech has become a major focus in speech science and technology. As dialogue systems evolve from transactional tools into socially interactive agents, they demand increasingly accurate automatic speech recognition (ASR). At the same time, conversational data offers unique insights into human speech processing. Drawing on the cross-layer optimization principle from communications engineering, I adopt a similar view of how meaning is accessed across multiple levels of speech information. In this talk, I present findings from my group’s work on integrating pronunciation and prosodic variation into ASR for conversational speech. Our hybrid approach—combining data-driven and knowledge-based methods—proves especially effective in low-resource settings. While transformer-based models often outperform classical systems, the latter still excel with short, fragmented utterances when paired with linguistic knowledge. Beyond ASR, our methods inform fields like pathological speech analysis, dementia prediction, and assistive speech technologies.

Bio: Barbara Schuppler studied Physics and Spanish Philology at the University of Graz and the Universidad Autónoma de Madrid, completing a diploma thesis in experimental physics in 2007. She conducted her dissertation within the Marie-Curie RTN "Sound-to-Sense" at Radboud University Nijmegen, with research visits at NTNU Trondheim. After working as teacher at the Graz International Bilingual School, she was awarded an FWF Hertha-Firnberg Grant in 2012 and joined the Signal Processing and Speech Communication Laboratory at TU Graz. Now Associate Professor at TU Graz, her research interests include the investigation of methods for quantitative analyses of prosody and pronunciation variation in conversational speech, the integration of gained phonetic and linguistic knowledge into speech technology, with a specific focus on applications in the educational and healthcare sector.

[camera] JSALT 2025 - Plenary Talk - B. Schuppler: Cross-layer Models for Low-Res Conversational ASR

Load More... Subscribe

Johns Hopkins University, Whiting School of Engineering

Center for Language and Speech Processing
Hackerman 226
3400 North Charles Street, Baltimore, MD 21218-2680

(410) 516-4237
[email protected]

Whiting School of Engineering
Johns Hopkins University

Privacy Statement
Legal

© 2025 Johns Hopkins University

Center for Language and Speech Processing