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

Seminars

Categories

Seminars

Student Seminars

Workshops
Tags
2016 2016 December Mesgarani 2017 2017 Bansal May 2017 Chelba April 2017 Jamieson April 2017 March Gannot 2017 Satheesh April 2017 Stevenson April 2018 2019 2020 2021 2022 2023 Abend Anastasopoulos Anumanchipalli April April Foreman Auli Aziz Balasubramanian Bamman Barker Belyy Bhati Biadsy and Kanevsky Blasi Bonial Boyd-Graber Brian Roark Broderick Brummer Center for Language and Speech Processing Chen Cherry Chiang Chintala Chiori Hori Cho Choi Christensen Chu-Carroll CLSP CMU Cohen Computer Science Cristia December Deep Learning Diab Dietz Don P. Giddens Inaugural Professorial Lecture Dredze Ellis Ettinger Fan February Februray Field Foulds Frederico Frermann Fried Fyshe Gabriel Gannot Gao Gardner Georgiou Goharian Gong Google AI Google Research Gordon Gorman Graca Grant Green Guan Guillermo Cecchi Gurevych Hajishirzi Hakkani-Tur Hansen Harrigian Harrison Hasegawa-Johnson He Heafield Heald Hershey Hoffman Hopcroft Hovy Huang Hutter Hwang Jacek Mostwin January Jason Eisner Ji Johns Hopkins University Joshi Kamper Kann Karakos Kayser Khashabi Kim Kirov Knill Kriz Kyunghyun Cho Lapata Lee Levy Li Linzen Lipton Litman Louis Lozano-Diez Marasovic March Markus Muller Mathias May McCoy Mekyska Metze Mi Mihalcea Miloš Cerňak Mimno Minematsu Mitchell Mitra Mohamed Morency Moreno Mower-Provost Mysore Nakov Natural Language Modeling Natural Language Processing Neural Machine Translation New York University November October Ostendorf Ouyang Parada Peng Peterson Pinter Piper Prabhavalkar Pustejovsky Radinsky Ramabhadran Rasooli Rastrow Re Rebecca Resnik Rene Vidal Ringger Romanization Rose Rudinger Rush Saenko Sainath Schneider September Sha Shinji Watanabe Singh Siskind Sisman Slaney Smaragdis Smith Socher Solorio Song Sproat Stent Steorts Suhr Summer School Sun Szymanski Third Frederick Jelinek Memorial Summer Workshop Ture Viertanen Wang White Whiting School of Engineering ws12plenary ws14 ws14clamr ws14mt ws14participant ws14prelim ws14semantics ws16plenary Wu Xu Yang Yarmohammadi Yatskar Youn Yulia Tsvetkov Zelasko Zeldes Zellers Zettlemoyer Zhao Zhou Zhu

2022 Jan February 2023 Mar 2024

Sun	Mon	Tue	Wed	Thu	Fri	Sat
			1	2	3 Sasha Rush (Cornell University) “Pretraining Without Attention” 12:00 pm Sasha Rush (Cornell University) “Pretraining Without Attention” @ Hackerman Hall B17 Feb 3 @ 12:00 pm – 1:15 pm Tickets Abstract Transformers are essential to pretraining. As we approach 5 years of BERT, the connection between attention as architecture and transfer learning remains key to this central thread in NLP. Other architectures such as CNNs and RNNs[...]	4
5	6	7	8	9	10	11
12	13	14	15	16	17	18
19	20	21	22	23	24	25
26	27	28

2022 Jan February 2023 Mar 2024

Subscribe to filtered calendar

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

Seminars

Seminar Calendar
Seminar Videos

Subscribe to the CLSP Seminars Mailing List

RT @suzyjstyles: #Multilingual #LangDev + #SpeechProc friends! We’ve prepped >30hrs of our English/Mandarin code-switched speech for the @… 20 hours
RT @cheeesio: I'm on the job market! (industry/post-doc/faculty) I work on multilinguality and low-resource NLP, with a focus on computa… Feb 02
RT @HopkinsEngineer: [email protected]_Sulam, assistant professor in @JHUBME, has been named a recipient of the @NSF's Early CAREER Award for his proj… Feb 01
RT @Carankt: This is super interesting! I have been inclined towards exploring curriculum learning on LLMs and did some ground work on the… Jan 31
RT @amuuueller: Announcing the BabyLM 👶 Challenge, the shared task at @conll_conf and CMCL'23! We’re calling on researchers to pre-train… Jan 31
RT @leibnyPaola: #MerlionChallenge News: Can your speech processing systems handle the challenge of vibrantly multiling… Jan 30
RT @ranganr: Can we trust AI? Dr. Rama Chellappa says we can trust AI for its benefits but also one needs to be cautious and control it fo… Jan 28
RT @Shadowkiller331: #ICLR2023 #ICLR So happy to get my first ICLR paper, we propose a new textual defense pipeline (TextShield) against te… Jan 23
RT @MuyinatuBell: The 2023 @NSF REU Program in Computational Sensing and Medical Robotics @JohnsHopkins is accepting applications! More de… Jan 16
RT @suchisaria: Who are your favorite people working in Health AI? I mean those who understand ML and, more importantly, how to bring these… Jan 16
RT @ieee_slt: Day 2 wrapped up with a keynote by @verena_rieser and a stimulating poster session. Follow us for updates on Day 3, which fe… Jan 12
RT @SNFAgoraJHU: The SNF Agora Institute is currently welcoming applications for our 2023-2024 visiting fellows cohort. Applications will b… Jan 12
RT @mdredze: TLDR: we do a lot of research at @JohnsHopkins Jan 12
RT @mdredze: @percyliang We have some postdoc positions at @jhuclsp if they are interested: https://t.co/YaBQyocEHj Jan 12

Seminar Videos

October 23, 2020

Abstract

Deep learning has been successfully applied to solve various complex problems. However, deep learning algorithms have also been employed to create deepfakes (e.g. fake audios, images and videos) for misleading information. Deepfakes are increasingly detrimental to privacy, society security and even democracy. This talk first introduces the state-of-the-art voice generation algorithms which can be leveraged to manipulate or create audio audio content. Following by an overview of the audio deepfake detection efforts done by the research community through the ASVspoof challenges, the talk concludes with a discussion of the challenges in audio deepfake detection.

Biography

Zhizheng Wu is a research scientist at Facebook. He pioneered in speaker verification spoofing detection research and initiated the first Speaker Verification Spoofing and Countermeasures (ASVspoof) challenge, which becomes a biennial event. Zhizheng also co-organized the first Voice Conversion Challenge (VCC 2016) and organized the Blizzard Challenge 2019 edition. He is also the creator of the SAS corpus which established the protocol for ASVspoof research and the creator of Merlin, which is an open-source neural network-based speech synthesis toolkit.

Audio Deepfake Detection: An Overview and its Challenges -- Zhizheng Wu (Facebook)

Absract

In this talk, you will get exposure to the various types of deep learning frameworks – declarative and imperative frameworks such as TensorFlow and PyTorch. After a broad overview of frameworks, you will be introduced to the PyTorch framework in more detail. We will discuss your perspective as a researcher and a user, formalizing the needs of research workflows (covering data pre-processing and loading, model building, etc.). Then, we shall see how the different features of PyTorch map to helping you with these workflows.

Biography

Soumith Chintala is a Researcher at Facebook AI Research. He works on deep learning and high-performance computing.

An Overview of Deep Learning Frameworks and an Introduction to PyTorch -- Soumith Chintala - 2017

$Abstract Generating usable internal representations of speech input requires, among other operations, fractionating the signals into temporal units/chunks of the appropriate granularity. Adopting (and adapting) Marr’s (1982) approach to vision, a perspective is outlined that formulates linking hypotheses between specific neurobiological mechanisms (for example cortical oscillations and phase-locking) and the representations that underlie auditory cognition (for example syllables). Focusing on the implementational and algorithmic levels of description, I argue that the perception of sound patterns requires a multi-time resolution analysis. In particular, recent experimental data from psychophysics, MEG (Luo & Poeppel 2007), and concurrent EEG/fMRI (Giraud et al. 2007) suggest that there exist two privileged time scales that form the basis for constructing elementary auditory percepts. These ‘temporal primitives’ permit the construction of the internal representations that mediate the analysis of speech and other acoustic signals. Biography David Poeppel is Professor in the Department of Biology, the Department of Linguistics, and the Neuroscience and Cognitive Science Program at the University of Maryland College Park and the Department of Psychology at New York University. Trained in neurophysiology, cognitive science, and cognitive neuroscience at MIT and UCSF, his lab is focused on the cognitive neuroscience of hearing, speech, and language. Although the lab uses all kinds of techniques, one principal methodology is magnetoencephalography (MEG).$

Temporal primitives in auditory cognition and speech perception – David Poeppel (UMD) - 2008

Probabilistic Methods for Classification - Gideon Mann - 2009

ABSTRACT:

This is an updated version of a talk given at ASRU 2013 in Olomouc.

Automatic pattern classifiers that output soft, probabilistic classifications—rather than hard decisions—can be more widely and more profitably applied, provided the probabilistic output is
well-calibrated. In the fields of automatic speaker recognition and automatic spoken language recognition, the regular NIST technology evaluations have placed a strong emphasis on cost effective
application and therefore on calibration. This talk will describe calibration solutions for these technologies, with emphasis on criteria for measuring the goodness of calibration—if we can measure
it, we can also optimize it.

The core of the talk is a derivation and a re-interpretation of cross-entropy, which is the standard objective function in machine learning for the supervised training of classifiers. The main
theoretical result is that cross-entropy represents the expected cost of making minimum-expected-cost Bayes decisions, based on the outputs of a softmax classifier. For this equivalence we use a special misclassification cost function, defined over a smooth range of cost values. In practice this means that classifiers trained with cross-entropy can be expected to work well over a wide range of different applications.

BIOGRAPHY:

Niko Brummer received B.Eng (1986), M.Eng (1988) and Ph.D. (2010) degrees, all in electronic engineering, from Stellenbosch University. He worked as researcher at DataFusion (later called Spescom DataVoice) and is currently chief scientist at AGNITIO. Most of his research for the last two decades has been applied to automatic speaker and language recognition and he has been participating in most of the NIST SRE and LRE evaluations in these technologies, from the year 2000 to the present. He has been contributing to the Odyssey Workshop series since 2001 and was organizer of Odyssey 2008 in Stellenbosch. His FoCal Toolkit is widely used for fusion and calibration in speaker and language recognition research.

His research interests include development of new algorithms for speaker and language recognition, as well as evaluation methodologies for these technologies. In both cases, his emphasis is on
probabilistic modelling. He has worked with both generative (eigenchannel, JFA, i-vector PLDA) and discriminative (system fusion, discriminative JFA and PLDA) recognizers. In evaluation, his focus is on judging the goodness of classifiers that produce probabilistic outputs in the form of well calibrated class likelihoods.

Binary and Multiclass Calibration in Speaker and Language Recognition - Niko Brummer (AGNITIO)

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

© 2023 Johns Hopkins University

Center for Language and Speech Processing