Dimitris Bertsimas

@StopAntisemites Ed Roberts

What's it like being an MIT Jewish student since 10/7? Get ready to have your jaw drop. Oct. 8: not even 24 hours passes since the worst attack on Jews occurs since the Holocaust does MIT's Coalition Against Apartheid (CAA) blames the massacre on ... Israel. No one is reprimanded. Oct. 13: CAA pushes for a “global day of Jihad”. No one is reprimanded. Oct. 19: students filmed chanting “One Solution: Intifada Revolution” & “From the river to the sea, Palestine will be free.” No one is reprimanded. Oct. 23: protesters disrupt classes, waving Palestinian flags & accusing MIT, Israel, and the U.S. of “genocide”. No one is reprimanded. Nov. 2: CAA members with a bullhorn and a drum roam through campus shouting anti-Israel slogans, barging into the President's office; they are escorted by the MIT police. No one is reprimanded. Nov 7: DEI staff member tells a Jewish student "they are not a protected class". DEI staff member Sophia Hasenfus even helped organize CAA rallies and openly endorsed statements justifying Hamas’s terror attack. Nov. 9 - CAA blockades Lobby 7, a major thoroughfare through campus. No one is reprimanded. Nov. 12 - an even larger group of protesters march across the Mass. Ave. bridge and gather at the Institute’s main entrance. No one is reprimanded. Nov. 13: MIT President Sally Kornbluth refuses to adopt the IHRA Definition of Antisemitism Nov. 16: the “Planetary Health: Indigenous Land, Peoples and Bodies" event takes place, led by an interfaith chaplain in which the Chaplain states Palestinians are being “wrongfully subjugated and oppressed by racist white European colonizers.” No one is reprimanded. Dec. 5: MIT President Sally Kornbluth miserably testifies in front of Congress. Dec. 6: a man harasses people at the Hillel building and urinates on it. No one is reprimanded. Dec. 13 - algorithms lecturer Mauricio Karchmer resigns due to MIT's failure of its Jewish students Dec. 21 - CAA offers support for Yemeni Houthi terrorists in a social media post. No one is reprimanded. Jan. 24 - campus wide hate initiative stresses Islamophobia comes out Feb. 12 - CAA again protests on campus and FINALLY gets a temporary suspension - 4 months after terrorizing Jews on campus. Feb. 18 - CAA crashed an MLK event. No one is reprimanded.

Dimitris Bertsimas, Georgios Margaritis: Global Optimization: A Machine Learning Approach arxiv.org/abs/2311.01742 arxiv.org/pdf/2311.01742

Interpretable machine learning methodologies are powerful tools to diagnose and remedy system-related bias in care, such as disparities in access to postinjury rehabilitation care. ja.ma/3P8Sdjc @hayfarani @dbertsim @AnthonyGebran @LMaurerMD

This paper would not have been possible without my coauthors @NeelNanda5, Matthew Pauly, Katherine Harvey, @mitroitskii, and @dbertsim or all the foundational and inspirational work from @ch402, @boknilev, and many others! Read the full paper: arxiv.org/abs/2305.01610

Precision and recall can also be helpful guides, and remind us that it should not be assumed a model will learn to represent features in an ontology convenient or familiar to humans.

While we found tons of interesting neurons with sparse probing, it requires careful follow up analysis to draw more rigorous conclusions. E.g., athlete neurons turn out to be more general sport neurons when analyzing max average activating tokens.

What happens with scale? We find representational sparsity increases on average, but different features obey different scaling dynamics. In particular, quantization and neuron splitting: features both emerge and split into finer grained features.

Results in toy models from @AnthropicAI and @ch402 suggest a potential mechanistic fingerprint of superposition: large MLP weight norms and negative biases. We find a striking drop in early layers in the Pythia models from @AiEleuther and @BlancheMinerva.

Early layers seem to use sparse combinations of neurons to represent many features in superposition. That is, using the activations of multiple polysemantic neurons to boost the signal of the true feature over all interfering features (here “social security” vs. adjacent bigrams)

But what if there are more features than there are neurons? This results in polysemantic neurons which fire for a large set of unrelated features. Here we show a single early layer neuron which activates for a large collection of unrelated n-grams.

Neural nets are often thought of as feature extractors. But what features are neurons in LLMs actually extracting? In our new paper, we leverage sparse probing to find out arxiv.org/abs/2305.01610. A 🧵:

One large family of neurons we find are “context” neurons, which activate only for tokens in a particular context (French, Python code, US patent documents, etc). When deleting these neurons the loss increases in the relevant context while leaving other contexts unaffected!

Reducing overall deaths and increase access for patients waiting for lung transplants. prnewswire.com/news-releases/…

As part of HIAS and together with Professor Georgios Stamou from NTUA, Greece we are offering a course on Universal AI (in English, free of charge) aicourse2023.hellenic-ias.org on July 3-5, 2023 in Athens, Greece. Prospective participants can declare their interest in the website.

Delighted to share that our paper "A new perspective on low-rank optimization" has just been accepted for publication by Math Programming! Valid & often strong lower bounds on low-rank problems via a generalization of the perspective reformulation from mixed-integer optimization

📢New preprint alert! arxiv.org/abs/2303.07695 We use sampling schemes and clustering to improve the scalability of deterministic Bender's decomposition on data-driven network design problems, while maintaining optimality. w/ @dbertsim, Jean Pauphilet, and Periklis Petridis

The paper presents a novel holistic deep learning framework that improves accuracy, robustness, sparsity, and stability over standard deep learning models, as demonstrated by extensive experiments on both tabular and image data sets. arxiv.org/abs/2110.15829

My book with David Gamarnik “Queueing Theory: Classical and Modern Methods” was published. It was a long journey that lasted two decades but both of us are delighted with the journeys completion. For more details see dynamic-ideas.com/books/quueing-…

The Holistic AI in Medicine (HAIM) framework from @dbertsim et al. in @AIHealthMIT is a pipeline to receive multimodal patient data + use generalizable pre-processing + #machinelearning modelling stages adaptable to multiple health related tasks. nature.com/articles/s4174…

If you are into #MachineLearning and #Statistics check this out. I would also highly recommend the Machine Learning under a modern optimization lens book by @dbertsim and Dunn. Here are two two teaser must watch imo youtube videos youtu.be/7w9aRrYgGEs youtu.be/jJgdJaCo568