Frühere Suchanfragen

Keine früheren Suchanfragen

Suchoptionen

Nur verfügbar, wenn angemeldet.
mastodontech.de ist einer von vielen unabhängigen Mastodon-Servern, mit dem du dich im Fediverse beteiligen kannst.
Offen für alle (über 16) und bereitgestellt von Markus'Blog

Verwaltet von:

Serverstatistik:

1,4 Tsd.
aktive Profile

mastodontech.de: ÜberStatusProfilverzeichnisDatenschutzerklärung

Mastodon: ÜberApp herunterladenTastenkombinationenQuellcode anzeigenv4.4.4

#neurips

0 Beiträge0 Beteiligte0 Beiträge heute
Stefan Siegert<p>How often have I heard the sentence "the optimiser got stuck in a local minimum" (myself included)? This is most likely wrong.</p><p>Dauphine et al (2014) [1] show that in high dimensional loss surfaces, local extrema are a rare exception and saddle points become exponentially more likely.</p><p>They use some cool results from statistical physics and random matrix theory, and also have numerical experiments.</p><p>[1] <a href="https://proceedings.neurips.cc/paper_files/paper/2014/file/04192426585542c54b96ba14445be996-Paper.pdf" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">proceedings.neurips.cc/paper_f</span><span class="invisible">iles/paper/2014/file/04192426585542c54b96ba14445be996-Paper.pdf</span></a></p><p><a href="https://mastodon.social/tags/mathematics" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>mathematics</span></a> <a href="https://mastodon.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machinelearning</span></a> <a href="https://mastodon.social/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a> <a href="https://mastodon.social/tags/neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips</span></a> <a href="https://mastodon.social/tags/paper" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>paper</span></a></p>
Helmholtz Imaging<p>✨ Thank you for being part of an inspiring year! Here’s to a bright 2025 w/new discoveries &amp; connections.</p><p>Holiday reads:<br>🔗 HI@ <a href="https://helmholtz.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a>: <a href="https://bit.ly/HI-at-NeurIPS-2024" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">bit.ly/HI-at-NeurIPS-2024</span><span class="invisible"></span></a> <br>🔗 Interview with Lena, HI Coordinator <span class="h-card" translate="no"><a href="https://helmholtz.social/@DKFZ" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>DKFZ</span></a></span>: <a href="https://bit.ly/41COnW7" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">bit.ly/41COnW7</span><span class="invisible"></span></a> </p><p>📸 Don't forget: Submit your best scientific images to our contest by 14/2: <a href="https://bit.ly/BSIC2025Call" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">bit.ly/BSIC2025Call</span><span class="invisible"></span></a></p><p><span class="h-card" translate="no"><a href="https://helmholtz.social/@association" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>association</span></a></span> <span class="h-card" translate="no"><a href="https://helmholtz.social/@DESYnews" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>DESYnews</span></a></span> <span class="h-card" translate="no"><a href="https://social.bund.de/@MDC_Berlin" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>MDC_Berlin</span></a></span> <span class="h-card" translate="no"><a href="https://bird.makeup/users/helmholtz_ai" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>helmholtz_ai</span></a></span> <span class="h-card" translate="no"><a href="https://helmholtz.social/@helmholtz_hmc" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>helmholtz_hmc</span></a></span></p>
rexi<p><a href="https://www.reuters.com/technology/artificial-intelligence/when-ai-vies-with-taylor-swift-hot-ticket-town-2024-12-16/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">reuters.com/technology/artific</span><span class="invisible">ial-intelligence/when-ai-vies-with-taylor-swift-hot-ticket-town-2024-12-16/</span></a></p><p>…nothing like the intimate affair it was decades ago, when a field of outliers could fit into a hotel bar. It has become fertile ground for <a href="https://mastodon.social/tags/corporations" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>corporations</span></a> to tout their wares and draw <a href="https://mastodon.social/tags/academics" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>academics</span></a> into newly lucrative business--The crowds were so massive that <a href="https://mastodon.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> began a day later than usual, so <a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/scientists" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>scientists</span></a> would not fight for hotel rooms the same night as a <a href="https://mastodon.social/tags/TaylorSwift" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>TaylorSwift</span></a> concert…</p>
Erika Varis Doggett<p>So I had gotten to the conference late and missed the rest of time award presentation and resulting mini talk from Ilya Sutskever, but a coworker linked me the video after. </p><p>And I gotta say, I really wish more people would think critically about WHAT they are actually trying to solve and WHY when they advocate for something like “superintelligence”. </p><p>Just…to think for a minute.</p><p><a href="https://mas.to/tags/neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips</span></a> <a href="https://mas.to/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mas.to/tags/ML" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ML</span></a></p>
Dirk Van den Poel<p>We have lift off at the <a href="https://mastodon.online/tags/NeurIPS2024" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS2024</span></a> Workshops in Vancouver, BC (Canada). I decided to focus on Adaptive Foundation Models. <a href="https://mastodon.online/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLM</span></a> <a href="https://mastodon.online/tags/largelanguagemodels" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>largelanguagemodels</span></a> <a href="https://mastodon.online/tags/finetuning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>finetuning</span></a> <a href="https://mastodon.online/tags/RAG" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RAG</span></a> <a href="https://mastodon.online/tags/RatrievalAugmentedGeneration" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RatrievalAugmentedGeneration</span></a> <a href="https://mastodon.online/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a></p>
Dirk Van den Poel<p>Heading to Vancouver, BC (Canada) for the <a href="https://mastodon.online/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> Conference to represent Ghent University’s Data Science for Business Program @NeurIPSConf <a href="https://mastodon.online/tags/GenAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GenAI</span></a> <a href="https://mastodon.online/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mastodon.online/tags/GenerativeAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GenerativeAI</span></a> <a href="https://mastodon.online/tags/orms" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>orms</span></a> <a href="https://mastodon.online/tags/DataScience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DataScience</span></a> <a href="https://mastodon.online/tags/DS4B" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DS4B</span></a> <a href="https://mastodon.online/tags/NeurIPS24" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS24</span></a></p>
Veronika Cheplygina<p>Anybody here going to be at <a href="https://dair-community.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> <a href="https://dair-community.social/tags/NeurIPS2024" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS2024</span></a>? Me and <span class="h-card" translate="no"><a href="https://social.itu.dk/@amelia" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>amelia</span></a></span> are going and we will mostly be at <a href="https://dair-community.social/tags/WiML" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>WiML</span></a> and affinity workshops, and the datasets track</p>
Chloé Azencott<p>Academia has a travel problem.</p><p>Attending faraway conferences is ecologically unsustainable, and creates a huge barrier to entry for people who are disabled, have care duties, or lack resources, among others.</p><p>One piece of the solution is to create local satellites of major conferences. This idea has been successfully implemented in Paris for <a href="https://lipn.info/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> for a few years, and I'm happy to have joined the advising committee of NeurIPS@Paris (Dec 4&amp;5 this year)</p><p>Check it out: <a href="https://neuripsinparis.github.io/neurips2024paris/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">neuripsinparis.github.io/neuri</span><span class="invisible">ps2024paris/</span></a></p>
aijobs.net => foorilla.com<p>HIRING: Founding AI Engineer, Agents / New York</p><p>👉 <a href="https://ai-jobs.net/J198951/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">ai-jobs.net/J198951/</span><span class="invisible"></span></a></p><p><a href="https://mstdn.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mstdn.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MachineLearning</span></a> <a href="https://mstdn.social/tags/DataJobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DataJobs</span></a> <a href="https://mstdn.social/tags/Jobsearch" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Jobsearch</span></a> <a href="https://mstdn.social/tags/MLjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MLjobs</span></a> <a href="https://mstdn.social/tags/bigdata" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>bigdata</span></a> <a href="https://mstdn.social/tags/DataScience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DataScience</span></a> <a href="https://mstdn.social/tags/AIjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AIjobs</span></a> <a href="https://mstdn.social/tags/techjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>techjobs</span></a> <a href="https://mstdn.social/tags/hiring" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>hiring</span></a> <a href="https://mstdn.social/tags/HiringAlert" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HiringAlert</span></a> <a href="https://mstdn.social/tags/Agents" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Agents</span></a> <a href="https://mstdn.social/tags/EngineerJobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>EngineerJobs</span></a> <a href="https://mstdn.social/tags/NYjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NYjobs</span></a> <a href="https://mstdn.social/tags/LLMs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLMs</span></a> <a href="https://mstdn.social/tags/python" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>python</span></a> <a href="https://mstdn.social/tags/ICLR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ICLR</span></a> <a href="https://mstdn.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> <a href="https://mstdn.social/tags/ICML" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ICML</span></a> <a href="https://mstdn.social/tags/CCVPR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>CCVPR</span></a></p>
aijobs.net => foorilla.com<p>HIRING: AI Engineer Intern, Agents / New York (Remote for exceptional candidates)</p><p>👉 <a href="https://ai-jobs.net/J198461/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">ai-jobs.net/J198461/</span><span class="invisible"></span></a></p><p><a href="https://mstdn.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mstdn.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MachineLearning</span></a> <a href="https://mstdn.social/tags/DataJobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DataJobs</span></a> <a href="https://mstdn.social/tags/Jobsearch" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Jobsearch</span></a> <a href="https://mstdn.social/tags/MLjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MLjobs</span></a> <a href="https://mstdn.social/tags/bigdata" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>bigdata</span></a> <a href="https://mstdn.social/tags/DataScience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DataScience</span></a> <a href="https://mstdn.social/tags/AIjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AIjobs</span></a> <a href="https://mstdn.social/tags/techjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>techjobs</span></a> <a href="https://mstdn.social/tags/hiring" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>hiring</span></a> <a href="https://mstdn.social/tags/Agents" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Agents</span></a> <a href="https://mstdn.social/tags/internship" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>internship</span></a> <a href="https://mstdn.social/tags/NYjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NYjobs</span></a> <a href="https://mstdn.social/tags/LLMs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLMs</span></a> <a href="https://mstdn.social/tags/python" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>python</span></a> <a href="https://mstdn.social/tags/ICLR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ICLR</span></a> <a href="https://mstdn.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> <a href="https://mstdn.social/tags/ICML" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ICML</span></a> <a href="https://mstdn.social/tags/CCVPR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>CCVPR</span></a></p>
Ian Holmes<p>My top three voice recognition errors from the <a href="https://mastodon.social/tags/neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips</span></a> live transcript (ML in structural biology workshop):</p><p>3. Kagglers =&gt; Cavaliers<br>2. AlphaFold =&gt; Alcohol<br>1. Generative models =&gt; Genital models</p><p><a href="https://mastodon.social/tags/neurips2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips2023</span></a></p>
Ian Holmes<p>Max Welling: “I’ve been long, long in denial that text based models could help you solve some physics problem… still actually believe that, but never mind. My uncertainty is getting bigger on that one” <a href="https://mastodon.social/tags/neurips2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips2023</span></a> <a href="https://mastodon.social/tags/neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips</span></a></p>
Ian Holmes<p>Chris Ré in <a href="https://mastodon.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> invited talk: "It's amazing what OpenAI has done. Ilya should get the Turing Award. Maybe not Employee Of The Month. Sorry, couldn't help myself" 😆 <a href="https://mastodon.social/tags/NeurIPS2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS2023</span></a></p>
Ian Holmes<p>This talk was actually one of my <a href="https://mastodon.social/tags/neurips2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips2023</span></a> highlights so far, explaining a recently discovered and quite interesting phenomenon that is relatively accessible to non specialists <a href="https://en.wikipedia.org/wiki/Double_descent" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">en.wikipedia.org/wiki/Double_d</span><span class="invisible">escent</span></a> <a href="https://mastodon.social/tags/neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips</span></a></p>
Leshem Choshen<p><a href="https://sigmoid.social/tags/neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips</span></a> keynote <br>(with my live jetlagged interpretation)<br>from <br><span class="h-card" translate="no"><a href="https://sigmoid.social/@StableDiffusion" class="u-url mention" rel="nofollow noopener" target="_blank">@<span>StableDiffusion</span></a></span><br> creator: <br>scaling is not the solution<br>A keynote to restart the debate <a href="https://sigmoid.social/tags/scalemodels" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>scalemodels</span></a> <br><a href="https://sigmoid.social/tags/LLMs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLMs</span></a> <a href="https://sigmoid.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MachineLearning</span></a> <a href="https://sigmoid.social/tags/GPTs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GPTs</span></a><br><a href="https://sigmoid.social/tags/ML" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ML</span></a> <a href="https://sigmoid.social/tags/NLP" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NLP</span></a> <a href="https://sigmoid.social/tags/nlproc" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>nlproc</span></a> <a href="https://sigmoid.social/tags/GPT" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GPT</span></a></p>
Ian Holmes<p>I am attending <a href="https://mastodon.social/tags/NeurIPS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>NeurIPS</span></a> for the first time, and in New Orleans for the first time. Which of these two things (<a href="https://mastodon.social/tags/neurips2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips2023</span></a> or NO) is the wilder? Sheer force of numbers suggests that the city has to outweigh the conference in chaos potential. And yet… this conference is really amazing. Genuinely stumped on this question (but will be mostly attending the conference and not hanging out in the French Quarter)</p>
Andrew Lampinen<p>Very excited to head to NeurIPS! Feel free to reach out if you want to chat about any of our recent work on LMs, agents, interpretability, representational alignment, etc. You can find me:</p><p>At the poster for our work on "Passive learning of active causal strategies in agents and language models"<br><a href="https://sigmoid.social/@lampinen/110434383859776741" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">sigmoid.social/@lampinen/11043</span><span class="invisible">4383859776741</span></a><br>Tue 12 Dec 5:15 p.m. CST — 7:15 p.m. CST<br>Great Hall &amp; Hall B1+B2 (level 1) #825<br><a href="https://nips.cc/virtual/2023/poster/72481" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">nips.cc/virtual/2023/poster/72</span><span class="invisible">481</span></a><br>1/3<br><a href="https://sigmoid.social/tags/Neurips" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Neurips</span></a> <a href="https://sigmoid.social/tags/neurips2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>neurips2023</span></a></p>
Cheng Soon Ong

Implicit gradients are closely related to fixed points, which have been used in neural network equilibrium models. We show in this paper that you can think of deep equilibrium models as maximum a posteriori estimates of some exponential family model.
arxiv.org/abs/2211.05943
The Gaussian special case retrieves score based models, which Russell will present at SBM Workshop at NeurIPS.
score-based-methods-workshop.g
#NewPaper #MachineLearning #NeurIPS #arxiv

arXiv.orgDeep equilibrium models as estimators for continuous latent variablesPrincipal Component Analysis (PCA) and its exponential family extensions have three components: observations, latents and parameters of a linear transformation. We consider a generalised setting where the canonical parameters of the exponential family are a nonlinear transformation of the latents. We show explicit relationships between particular neural network architectures and the corresponding statistical models. We find that deep equilibrium models -- a recently introduced class of implicit neural networks -- solve maximum a-posteriori (MAP) estimates for the latents and parameters of the transformation. Our analysis provides a systematic way to relate activation functions, dropout, and layer structure, to statistical assumptions about the observations, thus providing foundational principles for unsupervised DEQs. For hierarchical latents, individual neurons can be interpreted as nodes in a deep graphical model. Our DEQ feature maps are end-to-end differentiable, enabling fine-tuning for downstream tasks.
AngesagtLive-Feeds
Über