Alexandros G. Dimakis (Alex Dimakis) (original) (raw)

Home

Research

Papers

Courses

Activities

Fun

I am interested in Generative AI, Information Theory and Machine Learning.

Resume.

Google Scholar Profile.

News

• Co-founded BespokeLabs.ai a startup working on data curation and data centric AI. We are hiring.
• DeepSeek-R1 did not release its reasoning data. We are curating a reasoning dataset in OpenThoughts and training OpenThinker aiming to make the best open-source (open-weights and open-data) reasoning models.
• N. Raoof, L. Rout, G. Daras, S. Sanghavi, C. Caramanis, S. Shakkottai, A.G. Dimakis,
  "Infilling Score: A Pretraining Data Detection Algorithm for Large Language Models,"
  ICLR 2025, (Paper) (Code).
• S. Gadre, G. Smyrnis, V. Shankar, et al.
  "Language models scale reliably with over-training and on downstream tasks,"
  ICLR 2025,(Arxiv) (Code).
• A. Aali, G. Daras, B. Levac, S. Kumar, A. G. Dimakis, J. Tamir
  "Ambient Diffusion Posterior Sampling: Solving Inverse Problems with Diffusion Models Trained on Corrupted Data,"
  ICLR 2025, (Arxiv) (Code).
• Datacomp-LM released: DCLM. This is the largest public dataset for LLM training and includes 300T tokens, effective LLM per-training code based on Open_LM and 53 evaluations. Also we release truly open-source pretrained DCLM LMs of sizes up to 7B with excellent performance, full training data and source code.
• D.J. Diaz, C. Gong, J.Ouyang-Zhang, J.M. Loy, J. Wells, D. Yang, A. D. Ellington, A. G. Dimakis and A. R. Klivans
  Stability Oracle: a structure-based graph-transformer framework for identifying stabilizing mutations
  Nature Communications
  bioarxiv
• Four papers accepted to NeurIPS 2023
  • S. Gadre, G. Ilharco, A. Fang, J. Hayase, G. Smyrnis, T. Nguyen, R. Marten, M. Wortsman, D. Ghosh, J. Zhang, E. Orgad,
    R. Entezari, G. Daras, S. M. Pratt, V. Ramanujan, Y. Bitton, K. Marathe, S. Mussmann, R. Vencu, M. Cherti, R. Krishna,
    P. W. Koh, O. Saukh, A. Ratner, S. Song, H. Hajishirzi, A. Farhadi, R. Beaumont, S. Oh, A.G. Dimakis,
    J. Jitsev, Y. Carmon, V. Shankar, L. Schmidt,
    “DataComp: In search of the next generation of multimodal datasets”
    Proc. of Neural Information Processing Systems (NeurIPS), Dec. 2023. Datasets and Benchmarks Track,
    Selected for Oral presentation.
    https://www.datacomp.ai/ (Project Page)
• • G. Daras, K. Shah, Y. Dagan, A. Gollakota, A. G. Dimakis, A. R. Klivans,
    “Ambient Diffusion: Learning Clean Distributions from Corrupted Data,”
    Proc. of Neural Information Processing Systems (NeurIPS), Dec. 2023.
    (Project Page) (Arxiv),(Code).
• • L. Rout, N. Raoof, G. Daras, C. Caramanis, A. G. Dimakis, S. Shakkottai,
    “Solving Inverse Problems Provably via Posterior Sampling with Latent Diffusion Models,”
    Proc. of Neural Information Processing Systems (NeurIPS), Dec. 2023.(Project Page),(Arxiv),(Code)
• • G. Daras, Y. Dagan, A.G. Dimakis, C. Daskalakis,
    “Martingale Diffusion Models: Mitigating Sampling Drift by Learning to be Consistent,
    ” Proc. of Neural Information Processing Systems (NeurIPS), Dec. 2023.(Project Page)
• Our paper on deep learning for time-series modeling used for electronic system design and electromagnetic (EM) interconnect analysis for signal integrity, accepted to ICCAD:
  • S. Ravula, V. Gorti, B. Deng, S. Chakraborty, J. Pingenot, B. Mutnury, D. Wallace, D. Winterberg, A. R. Klivans, A. G. Dimakis,
    “One-Dimensional Deep Image Prior for Curve Fitting of S-Parameters from Electromagnetic Solvers,”
    2023 International Conference on Computer-Aided Design (ICCAD 2023)(Project Page)
• T. Chen, C. Gong, D. J. Diaz, X. Chen, J. T. Wells, Q. Liu, Z. Wang, A. D. Ellington, A.G. Dimakis, A. R. Klivans,
  “HotProtein: A Novel Framework for Protein Thermostability Prediction and Editing,”
  Proceedings on the International Conference on Representation Learning (ICLR 2023).
• Elected IEEE Fellow for contributions to distributed coding and learning, 2022.
• Selected as Commissioner. Artificial Intelligence Commission on Competition, Inclusion, and Innovation
  by the US Chamber of Commerce to provide a roadmap for tech leadership to US policy makers.
• Faculty of the year award (for 2022). MS Program in Information Technology Management, (Voted by students).
• Keynote speaker, 14th IEEE Image and Multidimensional Signal Processing Workshop (IVMSP) 2022.
• Plenary speaker, 13th International Conference on the Image, 2022.
• Best Paper Award at UAI 2021 Workshop on Tractable Probabilistic Modeling.
• Recent talk: Generative models are the new sparsity
  Recent Berkeley Simons talk on deep generative models and how they can be used to solve inverse problems including Denoising, Missing data, Compressed Sensing and MRI.
• G. Daras, N. Raoof, Z. Gkalitsiou and A.G. Dimakis
  ‘‘Multitasking Models are Robust to Structural Failure: A Neural Model for Bilingual Cognitive Reserve,’’
  Proc. of Neural Information Processing Systems (NeurIPS), Dec. 2022.
  (Project Page) arxiv
• M. Jordan, J. Hayase, A. G. Dimakis, S. Oh
  ‘‘Zonotope Domains for Lagrangian Neural Network Verification,’’
  Proc. of Neural Information Processing Systems (NeurIPS), Dec. 2022.
  (Arxiv)
• G. Daras, Y. Dagan, A. G. Dimakis, C. Daskalakis
  ‘‘Score-Guided Intermediate Level Optimization: Fast Langevin Mixing for Inverse Problems.’’
  International Conference on Machine Learning (ICML), 2022.
  (Project Page)
• J. Whang, M. Delbracio, H. Talebi, C. Saharia, A. G. Dimakis, P. Milanfar,
  ‘‘Deblurring via Stochastic Refinement.’’
  Computer Vision and Pattern Recognition (CVPR) June 2022. (Oral Presentation)(Arxiv)
• Two papers accepted to NeurIPS 2021
  • Ajil Jalal, Marius Arvinte, Giannis Daras, Eric Price, Alexandros G. Dimakis and Jonathan I. Tamir,
    Robust Compressed Sensing MRI with Deep Generative Priors,
    (Project Page), (Arxiv)
  • Sriram Ravula, Georgios Smyrnis ,Matt Jordan, Alexandros G. Dimakis
    Inverse Problems Leveraging Pre-trained Contrastive Representations
    (Project Page) (Arxiv)
• Six papers accepted to ICML 2021
  • Ajil Jalal,Sushrut Karmalkar, A.G. Dimakis and E. Price,
    Instance-Optimal Compressed Sensing via Posterior Sampling,
    (Project Page)
  • Ajil Jalal,Jessica Hoffmann,Sushrut Karmalkar, A.G. Dimakis and E. Price,
    Fairness for Image Generation with Uncertain Sensitive Attributes
    (Project page)
  • Giannis Daras, Joseph Dean, Ajil Jalal, and A.G. Dimakis,
    Intermediate Layer Optimization for Inverse Problems using Deep Generative Models
    (Project Page)
  • Matt Jordan, A.G. Dimakis,
    Provable Lipschitz Certification for Generative Models
    (Arxiv), (Code)
  • Jay Whang,Qi Lei, A.G. Dimakis
    Solving Inverse Problems with a Flow-based Noise Model
    (Arxiv)
  • Jay Whang, Erik Lindgren, A.G. Dimakis,
    Composing Normalizing Flows for Inverse Problems
    (Arxiv)
    Best Paper Award at UAI 2021 Workshop on Tractable Probabilistic Modeling
• Congratulations to Qi Lei for winning theOden Institute Outstanding Dissertation Award for her awesome Phd dissertation
• Recent service:
  • TPC chair for MLSys 2021
  • AAAI 2021 Area chair, NeurIPS 2020 Area chair
• New paper: Intermediate Layer Optimization for Inverse Problems using Deep Generative Models
  by G. Daras, J. Dean, A. Jalal and A.G. Dimakis.(Code) (Paper)
• Four papers accepted to NeurIPS 2020
  • A. Jalal, L. Liu, A.G. Dimakis and C. Caramanis,
    Robust compressed sensing of generative models, (arxiv)
  • M. Jordan and A.G. Dimakis,
    Exactly Computing the Local Lipschitz Constant of ReLU Networks,(arxiv)
  • I. Daras, N. Kitaev, A. Odena and A.G. Dimakis
    SMYRF - Efficient attention using asymmetric clustering(arxiv)
  • M. Kocaoglu, S. Shakkottai, A.G. Dimakis, C. Caramanis and S. Vishwanath
    Applications of Common Entropy in Causal Inference(pdf)
• New Survey: Deep Learning Techniques for Inverse Problems in Imaging
  G. Ongie, A. Jalal, C. A. Metzler, R. G. Baraniuk, A. G. Dimakis and R. Willett
  Journal on Selected Areas in Information Theory, May 2020. (arxiv), ieeeXplore
• Your Local GAN: Designing Two Dimensional Local Attention Mechanisms for Generative Models
  Giannis Daras, Augustus Odena, Han Zhang, Alexandros G. Dimakis, CVPR 2020.
  (arxiv) | Code | Colab Notebook | Twitter Thread
• Deep Generative models and Inverse Problems, talk at 2019 Texas AI summit
  (Slides.pdf) (Slides.pptx)
  Related Video:GANs and Compressed Sensing talk
• Gradient Coding from Cyclic MDS Codes and Expander Graphs
  N. Raviv, I. Tamo, R. Tandon and A. G. Dimakis, IEEE Transactions on Information Theory
  (arxiv)
• Five papers accepted to NeurIPS 2019
  • Matt Jordan , Justin Lewis, Alexandros G. Dimakis
    Provable Certificates for Adversarial Examples: Fitting a Ball in the Union of Polytopes
    (pdf)
• • Qi Lei, Ajil Jalal, Inderjit S. Dhillon, and Alexandros G. Dimakis
    Inverting Deep Generative models, One layer at a time.
    (pdf)
• • Qi Lei, Jiacheng Zhuo, Constantine Caramanis, Inderjit S Dhillon, Alexandros G Dimakis
    Primal-Dual Block Frank-Wolfe
    (pdf)
• • Shanshan Wu, Sujay Sanghavi, Alexandros G. Dimakis
    Sparse Logistic Regression Learns All Discrete Pairwise Graphical Models (Spotlight)
    (pdf)
• • Shanshan Wu, Alexandros G. Dimakis, Sujay Sanghavi
    Learning Distributions Generated by One-Layer ReLU Networks
    (pdf)
• SysML 2019 paper on adversarial attacks on text classifiers
  • Q. Lei, L. Wu, P. Chen, A.G. Dimakis, I.S. Dhillon and M. Witbrock,
    Discrete Adversarial Attacks and Submodular Optimization with Applications to Text Classification
    Systems and Machine Learning (SysML), April 2019. (pdf) slides code
  • Press coverage: Nature News, VentureBeat, TechTalks
• ICML 2019 paper on learned compressed sensing matrices
  • S. Wu, A.G. Dimakis, S. Sanghavi, F.X. Yu, D. Holtmann-Rice, D. Storcheus, A. Rostamizadeh, S. Kumar,
    Learning a Compressed Sensing Measurement Matrix via Gradient Unrolling
    International Conference on Machine Learning (ICML), 2019. (slides) (pdf) (Tensorflow Code)
• New Paper accepted to NeurIPS 2018
  • Erik Lindgren, Murat Kocaoglu, A.G. Dimakis and Sriram Vishwanath
    Experimental Design for Cost-Aware Learning of Causal Graphs
    Neural Information Processing Systems, 2019. (pdf) (Short video ft. Erik's radio voice)
• Preprint
  • D. Van Veen, A. Jalal, E. Price, S. Vishwanath, and A.G. Dimakis
    Compressed Sensing with Deep Image Prior and Learned Regularization
    https:arxiv.org_abs_1806.06438 (arxiv) (Code)
• Preprint
  • The Robust Manifold Defense: Adversarial Training using Generative Models
    A. Ilyas, A. Jalal, E. Asteri, C. Daskalakis, A. G. Dimakis
    (arxiv)
• Paper accepted to Annals of Statistics
  • E. R. Elenberg, R. Khanna, A.G. Dimakis, and S. Negahban
    ‘‘Restricted Strong Convexity Implies Weak Submodularity’’,
    to appear in Annals of Statistics, 2018.
    (preprint)
• Teaching resources for Data Science and Machine learning
• Honored to receive the James Massey award
• NeurIPS, ICML, AISTATS Area chair
• Compressed Sensing using Generative models Pre-trained models see also GitHub CSGM
• ICLR 2018
  • A. Bora, E. Price, A.G. Dimakis
    AmbientGAN: Generative models from lossy measurements
    (Oral Presentation) (openreview), (code)
  • M. Kocaoglu, C. Snyder, A.G. Dimakis and S. Vishwanath
    CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training
    (arxiv) (code)
• NIPS 2017:
  • E.R. Elenberg, A.G. Dimakis, M. Feldman, and A. Karbasi
    ‘‘Streaming Weak Submodularity: Interpreting Neural Networks on the Fly’’,
    (NIPS), 2017. (Oral presentation)(arxiv) code
  • Model-Powered Conditional Independence Test R. Sen, A.T. Suresh, K. Shanmugam, A.G. Dimakis and S. Shakkottai (NIPS), 2017.
• Preprint
  • Gradient Coding from Cyclic MDS Codes and Expander Graphs
    N Raviv, I Tamo, R Tandon, AG Dimakis (arxiv)
• Six papers accepted to ICML 2017:
  • Compressed Sensing using Generative Models (arxiv) Code and Demo
    A. Bora, A. Jalal, E. Price, A. G. Dimakis
  • Gradient Coding (pdf)
    R. Tandon, Q. Lei, A. G. Dimakis, N. Karampatziakis.(Slides from ITA)
  • On Approximation Guarantees for Greedy Low Rank Optimization(arxiv)
    R. Khanna, E. Elenberg, A. G. Dimakis, S. Negahban.
  • Exact MAP Inference by Avoiding Fractional Vertices(arxiv)
    E. M. Lindgren, A. G. Dimakis, A. Klivans.
  • Cost-Optimal Learning of Causal Graphs (arxiv)
    M. Kocaoglu, A. G. Dimakis, S. Vishwanath.
  • Identifying Best Interventions through Online Importance Sampling (arxiv)
    R. Sen, K. Shanmugam, A. G. Dimakis and S. Shakkottai.
• Two papers accepted to ISIT 2017:
  • Entropic Causality and Greedy Minimum Entropy Coupling(arxiv)
    M. Kocaoglu, A. G. Dimakis, S. Vishwanath and B. Hassibi.
  • Coded Caching with Linear Subpacketization is Possible using Ruzsa-Szeméredi Graphs.(arxiv)
    K. Shanmugam, A. M. Tulino and A. G. Dimakis.
• Recent work on decoding brain signals using interpretable features:
  • H. Yi, Z. Xie, R. Reetzke, A.G. Dimakis and B. Chandrasekaran.
    Vowel decoding from single-trial speech-evoked electrophysiological responses: A feature-based machine learning approach.
    Brain and Behavior. April 2017; (Open Access)
• Check the cool projects from our undergraduate Data Science Lab course
• Entropic Causality paper to appear in AAAI 2017. (Arxiv)
  More information
• Two papers accepted to AISTATS 2017:
  • Contextual Bandits with Latent Confounders: An NMF Approach(pdf)
    R. Sen, K. Shanmugam, M. Kocaoglu, A. G. Dimakis and S. Shakkottai.
  • Scalable Greedy Feature Selection via Weak Submodularity. (pdf)
    R. Khanna, E. Elenberg, A. G. Dimakis, S. Neghaban and J. Ghosh
• Slides and notes from GraphDay overview talk on graph analytics and machine learning (pdf)
• Upcoming talk at Canadian Workshop on Information Theory (CWIT)(CWIT link)
• Two papers accepted to NIPS 2016:
  • Leveraging Sparsity for Efficient Submodular Data Summarization
    E. Lindgren, S. Wu, A. G. Dimakis (pdf)
  • Single Pass PCA of Matrix Products S. Wu, S. Bhojanapalli, S. Sanghavi, A. G. Dimakis (pdf)
• Preprint: Restricted Strong Convexity Implies Weak Submodularity
  E. Elenberg, R. Khanna, A. G. Dimakis, S. Negahban(pdf)
• Distributed Estimation of Graph 4-profiles
  E. R. Elenberg, K. Shanmugam, M. Borokhovich, A. G. Dimakis.
  in Proc. International World Wide Web Conference (WWW), 2016(Arxiv)
• Bipartite Correlation Clustering: Maximizing Agreements
  M. Asteris, A. Kyrillidis, D. Papailiopoulos, A. G. Dimakis, AISTATS 2016 (pdf)
• Three papers accepted to NIPS 2015
  • Orthogonal NMF through Subspace Exploration
    M. Asteris D. Papailiopoulos A. G. Dimakis(pdf)
  • Sparse PCA via Bipartite Matchings
    M. Asteris D. Papailiopoulos A. Kyrillidis A. G. Dimakis(pdf)
  • Learning Causal Graphs with Small Interventions
    K. Shanmugam, M. Kocaoglu, A.G. Dimakis, S. Vishwanath(arxiv)
• Stay on path: PCA along graph paths M. Asteris A. Kyrillidis A. G. Dimakis H. Yi B. Chandrasekaran International Conference on Machine Learning (ICML), Lille, France, 2015,(pdf) (slides)
• E.R. Elenberg, K. Shanmugam, M. Borokhovich and A.G. Dimakis,
  Beyond Triangles: A Distributed Framework for Estimating 3-profiles of Large Graphs (KDD 2015, to appear)
• I. Mitliagkas, M. Borokhovich, A.G. Dimakis and C. Caramanis, FrogWild!-Fast PageRank Approximations on Graph Engines (to appear in VLDB 2015).
• Two papers accepted to ISIT 2015
  • On approximating the sum-rate for multiple unicastsK. Shanmugam, M. Asteris and A.G. Dimakis
  • Batch Codes through Dense Graphs with High GirthA.S. Rawat, Z. Song, A.G. Dimakis and A. Gal
• Two papers accepted in NIPS 2014.
  • Sparse Polynomial Learning and Graph Sketching (Oral)
    M. Kocaoglu, K. Shanmugam, A.G. Dimakis, A. Klivans
  • On the Information Theoretic Limits of Learning Ising Models
    R. Tandon, K. Shanmugam, P. Ravikumar, A.G. Dimakis
• Batch Codes through Dense Graphs without Short CyclesA.G. Dimakis, A. Gal, A.S. Rawat, Z. Song
• Invited talk on coding theory for distributed storage at Algebra Codes and Networks at Bordaux. Talk Slides.
• Two papers accepted in ICML 2014.
  • Nonnegative Sparse PCA with Provable Guarantees
    M. Asteris, D. Papailiopoulos, A.G. Dimakis
    ICML video
  • Finding Dense Subgraphs via Low-Rank Bilinear Optimization
    D. Papailiopoulos, I. Mitliagkas, A.G. Dimakis, C. Carmanis
    ICML video
• Three papers accepted in ISIT 2014.
  • Bounding Multiple Unicasts through Index Coding and Locally Repairable Codes
  • Graph Theory versus Minimum Rank for Index Coding
  • Locality and Availability in Distributed Storage
• Our Locally Repairable Storage codes featured on High Scalability, StorageMojo and TechCrunch.
• Online Milibo tutorial Erasure codes over Hadoop
• New paper: Sparse PCA through Low-rank Approximations (to appear in ICML 2013).video of Sparse PCA talk
• Our Xorbas Hadoop locally repairable codes paper will appear in VLDB 2013. Visit the Xorbas HDFS project homepage.
• I have moved to UT Austin. My USC homepage will no longer be updated.
• Received a Google Faculty Research Award. Support gratefully acknowledged.
• Our paper received the Communications Society & Information Theory Society Joint Paper Award.

I maintain the Distributed Storage Wiki, an online bibliography about theoretical problems in large-scale distributed storage systems. (Temprary offline, will need new host for this)