Nathaniel Hudson

Ph.D. Computer Scientist | Assistant Professor at @illinoistech

[Nathaniel Hudson @Nchudson95]

I am super proud of Jordan Pettyjohn, an undergraduate student I had the privilege of working with this past summer, for winning the Student Research Competition at the @Supercomputing conference! 🏆

His work studied ablation strategies for toxicity in #LLMs.

[Globus Labs @LabsGlobus]

Excited to share our latest work: "SoK: On Finding Common Ground in Loss Landscapes Using Deep Model Merging Techniques"! 🧠
https://t.co/XAG79rS3ij

By @arhamk1216, @toddknife, @Nchudson95, @Mansi__S, @dcgrzenda, @aswathy__ajith, Jordan Pettyjohn, @chard_kyle, and @ianfoster.

[Nathaniel Hudson @Nchudson95]

This is a thread on research our group at @UChicago recently made public on how to address the phenomenon of "memorized" data in large language models (#LLMs).

The preprint is public and a blog is available for a lighter read (see thread). Check it out! 📰#MachineLearning #AI

[Mansi Sakarvadia @Mansi__S]

1/🧵New Research on Language Models!
Language models (LMs) often "memorize" data, leading to privacy risks. This paper explores ways to reduce that!
Paper: https://t.co/OBtYz9mJON
Code: https://t.co/x0C5I77CG3
Blog: https://t.co/nA6AH5rnXV

[Nathaniel Hudson @Nchudson95]

Hallucinations in GPT models are a huge problem. A lot of this relates to the trust that we extend to these models. That said, this is a relevant read and an important story to follow.
https://t.co/KDgx48BwWV

[Globus Labs @LabsGlobus]

Our postdoc, Nathaniel Hudson (@Nchudson95), has been selected to participate in the 2024 CPS Rising Stars workshop at the University of Virginia. The acceptance rate was ~16%. Amazing job, Nathaniel!
https://t.co/sj15slurmy

[Nathaniel Hudson @Nchudson95]

My recent publication on QoS-aware placement and scheduling of Edge AI featured in Future Generation Computer Systems (@comp_science) is now publicly available.

From now until May 26 2024, the link below will give you access to the paper for free. 😄
https://t.co/Z1oyGQpD2i

[Globus Labs @LabsGlobus]

Check out our new paper surveying modern approaches to serving Trillion Parameter models, a part of our effort to create robust AI infrastructure for AI in Science: https://t.co/NjtzXVdXQ1. Congrats @Nchudson95, Greg Pauloski, @mbaughman123, Alok Kamatar, @Mansi__S, and others!

[Globus Labs @LabsGlobus]

Attention Lens (https://t.co/KOkEqQi5mJ) was accepted to the NEURIPS Attributing Model Behavior at Scale Workshop (https://t.co/Yt8lKxi6e8) @NeurIPSConf. Congratulations @Mansi__S, @arhamk1216, @aswathy__ajith, @dcgrzenda, @Nchudson95, @AndrBauer1, @chard_kyle, @ianfoster

[Globus Labs @LabsGlobus]

It was great to see everyone from the @ParslProject community at ParslFest! Check out the talks from @WardLT2, @yadunand, @Nchudson95, @ValHayot, Matt, Greg and many others on the ParslFest website: https://t.co/c8KgRBLGeD

[Omer Rana @omerfrana]

Our first RuralAI (https://t.co/BDRrRkJNuT) workshop led by Melanie Oii @waikato & Shaleeza Sohail @Uni_Newcastle international University networks are key to address global challenges. @chard_kyle @Nchudson95 @auluck_nitin @_akk5. Join the conversation: https://t.co/GRQNSVykh5

[Globus Labs @LabsGlobus]

and our illustrious post-doc scholar Nathaniel Hudson (@Nchudson95) will explore "Federated Learning on Globus Compute using FLoX." 🖥️🌐

[Globus Labs @LabsGlobus]

Nathaniel's (@Nchudson95) WF-IoT paper, entitled “Adversarial Predictions of Data Distributions Across Federated Internet-of-Things Devices”, in collaboration with members from SAGE got accepted in the main conference track at IEEE WF-IoT (https://t.co/Ond3yRm3FD)

[Globus Labs @LabsGlobus]

Congrats to @LabsGlobus postdoc @Nchudson95 for his seat on the on the technical program committee for the IEEE DCOS S-IoT 2024 in Abu Dhabi this fall! #IoT #distributedcomputing

[Jianshu Liu @jianshu_liu]

What an enriching summer research #internship at #globus @LabsGlobus @globus! Grateful for the opportunity to learn, grow and contribute. Thanks for cherished memories with my new friends. @WardLT2 @Nchudson95 A special shoutout to my mentor, Dr. Kyle Chard @chard_kyle

[UK GSACS @UK_GSACS]

Thank you to all of our members for a wonderful and fun year! Also, Congratulations to all of our GSACS graduates!
Here are some pics from our final event of the academic year!

[Globus Labs @LabsGlobus]

A Work-in-Progress paper on balancing federated learning trade-offs, authored by Ph.D. candidate Matt Baughman, postdoctoral scholar Nathaniel (@Nchudson95), et al., was awarded the Best WiP paper award at the 2023 @IEEEPerCom conference!

[Globus Labs @LabsGlobus]

@AndrBauer1 had his paper "Efficient Data Processing: Assessing the Performance of Different Programming Languages" and another paper co-authored with @Nchudson95 "Searching for the Ground Truth: Assessing the Similarity of Benchmarking Runs" accepted to the @ICPEconf. Congrats!

[Globus Labs @LabsGlobus]

@Nchudson95 will be travelling in style (with a travel grant!) to present his work on resource and performance tradeoffs in federated learning at the 2023 meeting of @ieeepervasive

#federatedlearning #ml #edgecomputing

[Globus Labs @LabsGlobus]

Matt Baughman and @Nchudson95 had their paper "Balancing Federated Learning Trade-Offs for Heterogeneous Environments" accepted to the Work in Progress session at @IEEEPerCom. Congrats! 🎉

[Nathaniel Hudson @Nchudson95]

@AustinClyde13 Congratulations, Austin!! You’re at the finish line. 🍾