Intersectional/Interdisciplinary ML research @datascicolumbia. Physics PhD. Passionate about responsible, empowering, ethical tech. Always stay humble and kind.
Happy to be in my favorite space at NeurIPS! To me, AI is political and my queerness is political and @QueerinAI holds space for me to embrace that rather than trying to suppress it. Beautiful discussions about community building in research and the world.
Next up, this afternoon I'll be on a panel "AI for All: Advancing Responsibility, Fairness, & Queer Representation in Technology" at @Hunter_College , hosted by my friend/colleague Raj Korpan. Such an important topic as building truly useful & trustworthy AI takes everyone!
My amazing colleague Susan McGregor is hiring a summer REU student to work on human-centered computing for digital news and media trust! Please share with any undergrads who might be interested!
Super excited that our paper on legislation to address bias in public sector algorithms was selected as one of the cover stories of MIT @SciPolRev! sciencepolicyreview.org/voluβ¦
Biggest congratulations to Dr. @GDezoort!! I canβt believe he just got his PhD because his work has been extraordinary for years already. Gage, youβre truly one of the most thoughtful and creative researchers I know. Canβt wait to see what you do next!
My talk 'AI as object of study: implications for science and society' from Monday's Responsible AI in the Natural Sciences at @Physics_CMU and @CMUBlockCenter is up on Youtube piped.video/uCd9UcbbVyc?t=3234 along with the other amazing talks and our interesting discussion panel!
Going back and forth from science spaces to ethics spaces is jarring. Today a speaker was discussing how they want a Chatgpt-like model for Boston dynamics robots that allow them to autonomously navigate & reason about the world. Isnβt that a cool interesting research problem? 1/
I am so tired of seeing stuff like this. Extraordinary claims require extraordinary proof and these overhyped unfounded statements do a disservice to everyone.
My body is so tired but my heart is SO full after getting to teach a whole 7 hour course about AI ethics for scientists. This is a topic I care so deeply about and Iβm incredibly grateful for the opportunity!
Super cool talk from @jonathanmukiibi about collecting radio data in Uganda to understand various aspects of the covid-19 pandemic. This is my favorite kind of project, understanding a unique, community specific need and developing meaningful collective solutions
Affinity workshop day is one of my favorite parts of NeurIPS. Starting the day off at @_LXAI workshop with Omar Florez talking about efficient temporal adaptation and machine unlearning for tweet data. Plus a call to the community to prioritize low resource approaches to AI!
Amazing talk from @dariogargas at the @BIRS_Math workshop on interoperability in AI. Really thankful to have spaces where we can meaningfully examine and critique trends in the field and discuss more sustainable and responsible solutions.
Super duper excited about this paper (maybe my favorite one I've done so far)! We explore the impact of equivariance in GNNs, and particularly on model efficiency, by isolating different architecture choices and measuring their impact of model accuracy AND our novel 'ant factor'
Fun day at the @APSphysics Congressional Visit talking to members of congress about R&D funding, investment in STEM education, dual intent visas for gradstudents, and more! Itβs nice to get out of my research bubble and be reminded of the impact basic science has in the world!
Happy tweet! Just found out one of my mentees whoβs summer research project I supervised was accepted early decision to Yale! Literally cried when she emailed me, Iβm so excited about the amazing things sheβll do & inspired by her thoughtfulness & dedication to community service
So excited to be teaching at the NYU AI School in January! If you are (or know) an undergrad in the NYC area interested in learning about ML/AI please apply! Open to folks without programming knowledge, please share widely!
Super excited to share that I've been promoted at Princeton! Officially I'm now a 'Associate Research Scholar', a position on the permanent research ladder. Looking forward to continuing my work at the intersection of ML, physics, and ethics with some relieved time pressure!
Anyone aware of good resources for philosophy learning? Iβve been feeling like I need to refresh and deepen my knowledge but Iβm not ready to commit to auditing a whole university class yet. Also if anyone is interested in a reading group hit me uppp
So inspired by researchers who take opportunities to acknowledge other folks in talks and other publicity! Yes! Hype your collaborators! Support your friends! Promote others all the time and democratize science!
Attended a really interesting talk from @StanfordHAI today on using self-supervised learning methods to create datasets to estimate progress towards the @UN sustainable development goals! (While waiting in line for my second vaccine dose. Yay science!)
Thrilled to discuss our work with a broad ML audience at the #MLSys2021 GNN workshop! Im biased π but particle physics data is such an cool area to apply geometric DL (we know so much about the geometry and expected symmetries) and hopefully now more folks think so too π
Excited for the MLSys GNN workshop today! Kicking off with a talk on GNNs in Biology/Medicine from @marinkazitnik. Really interesting to see how GNNs can be used for the full range of biological study: from diseases pathways to knowledge and patient modeling
The muon g-2 results are making me think about the differences between ML and particle physics publishing. Though the long timescales can be frustrating, the precision and rigor of the physics discovery process is one of my favorite things about the field! piped.video/watch?v=81PfYnpuβ¦
Our vCHEP paper characterizing Interaction Networks for particle tracking is up! We show that networks with as few as 6k parameters reach nearly perfect edge classification efficiency (eg more likely to fit on FPGAs π) @GDezoort@MarkSNeubauer@jmgduartearxiv.org/abs/2103.16701
Computing (or physics) isn't neutral just because it has a mathematical foundation. @theodorajewell does an excellent job demonstrating how technical tools can be used for digital (and physical!) colonization #apsmarch
Such a perfect talk for an AI Ethics discussion at @APSphysics March Meeting...physicists have a certain framework for understanding what 'water data' means, and @theodorajewell shows us how the broader context of this data
Every time I see @JasmineMcNealy speak about ecological data governance I'm more inspired by her! Data isn't static & systems that rely on data are multi-faceted and dynamic & effective regulation needs to recognize that data doesn't function only as personal property #apsmarch
Hi π€ I'm Savannah, currently in CT but getting back to NYC soon! I work at the intersection of machine learning (AI), physics, and tech ethics. These topics do critically intersect, I promise!
Our proof-of-concept paper using instance segmentation methods for particle tracking is up on arxiv! A work in progress, but I'm very excited to see where this work goes! arxiv.org/abs/2103.06509
. @marielpettee is defending her thesis today! So inspired to see such a high quality, impactful, and interdisciplinary body of work π guess Yale is the place for badass physics ML ladies, right @WonderMicky?
SO PROUD of my students! Itβs been so amazing to see them grow as data scientists and community members π this project focusing on helping a co-op deliver on its racial equity commitments by building a database of local minority owned businesses!
It's the last day of the data science training program I've been working with for the past several months and I had the awesome opportunity to talk about data ethics with them this morning. One of my students made this during my talk and I literally almost cried on camera
Networking with anxiety is so hard βΉοΈ like itβs hard enough sometimes for me to believe my best friends like me, my brain is like Sav, wtf would someone you donβt know weβll want to talk to you?