Sonal Joshi retweeted
I just finished the first year of my PhD. Rather than just share my own perspective, I sat down with 7 other students in my cohort and talked with them about their experiences in depth. In my new post, I share some of our biggest lessons: jonathanivey.github.io/2026/…
Sonal Joshi retweeted
I will present MedExpert, a clinician-annotated Factuality and Completeness dataset with Severity levels for Medical Chatbot Evaluation at ML4H in San Diego today. Happy to chat from Dec 1 to 3! See paper at: openreview.net/pdf?id=rkLAzD…
@mdredze @Alexir563 @jhuclsp @SonalSSJ
Sonal Joshi retweeted
We introduce WaltzRL🎶, a multi-agent RL framework that treats LLM safety as a positive-sum game between conversation & feedback agents. It strikes an elegant balance between helpfulness & harmlessness, boosting safety & reduces overrefusals without degrading capabilities!
💃New Multi-Agent RL Method: WaltzRL💃
📝: arxiv.org/abs/2510.08240
- Makes LLM safety a positive-sum game between a conversation & feedback agent
- At inference feedback is adaptive, used when needed
-> Improves safety & reduces overrefusals without degrading capabilities!
🧵1/5
Sonal Joshi retweeted
👩🔬International Day of Women and Girls in Science is this Saturday, #February11.
@UN_Women @WomenScienceDay
#WomenInScience #GirlsInScience #CdnSci
Es emocionante compartir historias de ciencia!👩🔬👩🔬
un.org/es/#un_podcasts
Sonal Joshi retweeted
#MerlionChallenge News:
Can your speech processing systems handle the challenge of vibrantly multilingual
naturalistic speech and figure out which language is spoken when?
Join the MERLion CCS challenge at #INTERSPEECH2023!
sites.google.com/view/merlio….
Sonal Joshi retweeted
In addition to our mentoring program, Prof Mark Dredze @mdredze will be hosting an admissions AMA on November 15 at 9:30 PM EST (to accommodate more time zones) 🙌.
Please pre-register to receive a zoom link: forms.gle/LTb2JrrRZnLa6CVM9
Reply with questions you want answered.
It's ✨grad app season✨ and @jhuclsp has a student-run application mentoring program for those looking for guidance with their application.
Open to prospective PhD applicants!
Apply here: forms.gle/MaMUfaEMr3URMMfS9
Submit materials by Nov 20.
Spread the word. #NLProc
Sonal Joshi retweeted
It's ✨grad app season✨ and @jhuclsp has a student-run application mentoring program for those looking for guidance with their application.
Open to prospective PhD applicants!
Apply here: forms.gle/MaMUfaEMr3URMMfS9
Submit materials by Nov 20.
Spread the word. #NLProc
Sonal Joshi retweeted
#Lhotse now supports annotating audio files with OpenAI's #Whisper! Here's a quick demo I created in 10 mins to transcribe the first 100 recordings from VoxCeleb.
See PR from @PiotrZelasko: github.com/lhotse-speech/lho…
Here's the sample audio: tinyurl.com/3vjxnubk
Sonal Joshi retweeted
Expert Session 8 - "Speech as a disease biomarker" by Catarina Botelho and Ayimnisagul Ablimi starts at 17:00! #ICASSP2022
Sonal Joshi retweeted
Join us for @ieeeICASSP tutorial 💬 🤖 "Adversarial Robustness & Reprogramming for Speech & Language Processing" I will present the 2nd section of the tutorial on neural reprogramming & prompting. Jointly with @pinyuchenTW @IBMResearch @GeorgiaTech_ECE 2022.ieeeicassp.org/tutorial…
Sonal Joshi retweeted
Just released Lhotse v1.2 with 4 new recipes, better consistency across recipe outputs (thanks @rdesh26) and a bunch of minor features and bug fixes.
github.com/lhotse-speech/lho…
Sonal Joshi retweeted
📢Submission portal is open for the 8th Doctoral Consortium at #INTERSPEECH2022📢
If you are a doctoral student👨💻and wishes to get feedback on your ongoing project from a panel of experts.
Submit your abstract before: 16 of June
More info on our website: isca-students.org/sacweb/
Sonal Joshi retweeted
📢Check out IEEE SPS's PROmotinG DiveRsity in Signal ProcESSing (PROGRESS) workshop. This workshop aims to prepare students and junior researchers (especially women and underrepresented minority groups) to pursue academic careers in signal processing. 👇
Thanks, GOOGLE, for sponsoring IEEE SPS PROGRESS Workshop.
I am pleased to announce that the registration for IEEE SPS PROGRESS is open now for students and postdocs.
The event is online. To receive a registration fee waiver, indicate it in your applic…lnkd.in/eJ5VJVzk
Sonal Joshi retweeted
Any #SpeechProc person who hasn't been living under a rock for the last few years has definitely come across the "transducer" model (formerly known as the RNN-Transducer). They look like this: (1/n)
Sonal Joshi retweeted
On that note, maybe I should do Twitter Bytes to summarize works in "Adversarial Attacks and Countermeasures" as these topics are not very well-known in the speech community? ♥️= upvote the idea
Sonal Joshi retweeted
Thanks @rdesh26 for tagging. Adversarial attacks cause havoc on both Speech and Speaker Recognition systems! Our team is working on how to protect against them and have many cool papers. Will follow up on details in the thread🧵
Sonal Joshi retweeted
``Adversarial Attacks and Defenses for Speaker Identification Systems. (arXiv:2101.08909v1 [eess.AS]),'' Sonal Joshi, Jesús Villalba, Piotr Żelasko, Laureano Moro-Velázquez, Najim Dehak, ift.tt/3phcsM6