About

Richard G. Baraniuk is a professor whose research focuses on signal processing, machine learning, and compressive sensing. His work involves developing innovative methods for data acquisition, analysis, and reconstruction, with applications across various scientific and engineering domains. Throughout his career, he has contributed significantly to the advancement of compressive sensing techniques and their practical implementations. Professor Baraniuk's background includes extensive research in signal processing and related fields, and he is actively involved in leading research groups and collaborative projects. His contributions have been recognized through numerous awards and fellowships, and he has mentored many students and postdoctoral researchers who have gone on to prominent academic and industry positions. His work continues to influence the development of efficient algorithms and systems for processing large-scale and high-dimensional data.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Biology
• Computational biology
• Computer Security
• Statistics
• Mathematics
• Data Mining
• Engineering
• Information Retrieval
• Data science
• Natural Language Processing
• Computer network
• Physics
• Seismology
• Genetics
• Algorithm
• Telecommunications
• Chromatography
• Geology
• Chemistry
• Distributed computing
• Biological system

Selected publications

• Beyond the Classroom Sample: Validating a STEM Biographical Data Measure with Online Adult Learners

  SSRN Electronic Journal · 2026-01-01

  preprintOpen accessSenior author
  PublisherDOI

• Minimizing Collateral Damage in Activation Steering

  ArXiv.org · 2026-05-01

  articleOpen accessSenior author

  Activation steering is a method for controlling Large Language Model (LLM) behavior by intervening in its internal representations to increase the alignment with a specific target feature direction. However, standard interventions, such as vector addition, often cause ``collateral damage", defined as unintended changes in the alignment of activations along other non-target feature directions. This damage occurs because standard methods implicitly assume the isotropy of non-target features. In this work, we provide a mathematical formalization of collateral damage and introduce a principled framework that models steering as a constrained optimization problem. Our method finds a new activation that minimizes the expected squared collateral change weighted by the empirical second-moment matrix of activations. This weighting encodes the nonuniform cost of the perturbation in different feature directions, in contrast to isotropic approaches that penalize changes uniformly in all feature directions. By accounting for the empirical second-moment of activations, our approach achieves more precise control while reducing the degradation of model performance on unrelated tasks.

  PublisherOA PDF

• CLEAR-3K: Assessing Causal Explanatory Capabilities in Language Models

  Underline Science Inc. · 2026-03-06

  otherOpen access

  We introduce CLEAR-3K, a dataset of 3,008 assertion-reasoning questions designed to evaluate whether language models can determine if one statement causally explains another. Each question presents an assertion-reason pair and challenge language models to distinguish between semantic relatedness and genuine causal explanatory relationships. Through comprehensive evaluation of 21 state-of-the-art language models (ranging from 0.5B to 72B parameters), we identify two fundamental findings. First, language models frequently confuse semantic similarity with causality, relying on lexical and semantic overlap instead of inferring actual causal explanatory relationships. Second, as parameter size increases, models tend to shift from being overly skeptical about causal relationships to being excessively permissive in accepting them. Despite this shift, performance measured by the Matthews Correlation Coefficient plateaus at just 0.55, even for the best-performing models. Hence, CLEAR-3K provides a crucial benchmark for developing and evaluating causal explanatory reasoning in language models, which is an essential capability for applications that require accurate assessment of causal relationships.

  PublisherDOI

• Leakage and Second-Order Dynamics Improve Hippocampal RNN Replay

  Open MIND · 2026-02-20

  preprintSenior author

  Biological neural networks (like the hippocampus) can internally generate "replay" resembling stimulus-driven activity. Recent computational models of replay use noisy recurrent neural networks (RNNs) trained to path-integrate. Replay in these networks has been described as Langevin sampling, but new modifiers of noisy RNN replay have surpassed this description. We re-examine noisy RNN replay as sampling to understand or improve it in three ways: (1) Under simple assumptions, we prove that the gradients replay activity should follow are time-varying and difficult to estimate, but readily motivate the use of hidden state leakage in RNNs for replay. (2) We confirm that hidden state adaptation (negative feedback) encourages exploration in replay, but show that it incurs non-Markov sampling that also slows replay. (3) We propose the first model of temporally compressed replay in noisy path-integrating RNNs through hidden state momentum, connect it to underdamped Langevin sampling, and show that, together with adaptation, it counters slowness while maintaining exploration. We verify our findings via path-integration of 2D triangular and T-maze paths and of high-dimensional paths of synthetic rat place cell activity.

  DOI

• Leakage and Second-Order Dynamics Improve Hippocampal RNN Replay

  ArXiv.org · 2026-02-20

  articleOpen accessSenior author

  Biological neural networks (like the hippocampus) can internally generate "replay" resembling stimulus-driven activity. Recent computational models of replay use noisy recurrent neural networks (RNNs) trained to path-integrate. Replay in these networks has been described as Langevin sampling, but new modifiers of noisy RNN replay have surpassed this description. We re-examine noisy RNN replay as sampling to understand or improve it in three ways: (1) Under simple assumptions, we prove that the gradients replay activity should follow are time-varying and difficult to estimate, but readily motivate the use of hidden state leakage in RNNs for replay. (2) We confirm that hidden state adaptation (negative feedback) encourages exploration in replay, but show that it incurs non-Markov sampling that also slows replay. (3) We propose the first model of temporally compressed replay in noisy path-integrating RNNs through hidden state momentum, connect it to underdamped Langevin sampling, and show that, together with adaptation, it counters slowness while maintaining exploration. We verify our findings via path-integration of 2D triangular and T-maze paths and of high-dimensional paths of synthetic rat place cell activity.

  PublisherOA PDF

• Minimizing Collateral Damage in Activation Steering

  arXiv (Cornell University) · 2026-05-01

  preprintOpen accessSenior author

  Activation steering is a method for controlling Large Language Model (LLM) behavior by intervening in its internal representations to increase the alignment with a specific target feature direction. However, standard interventions, such as vector addition, often cause ``collateral damage", defined as unintended changes in the alignment of activations along other non-target feature directions. This damage occurs because standard methods implicitly assume the isotropy of non-target features. In this work, we provide a mathematical formalization of collateral damage and introduce a principled framework that models steering as a constrained optimization problem. Our method finds a new activation that minimizes the expected squared collateral change weighted by the empirical second-moment matrix of activations. This weighting encodes the nonuniform cost of the perturbation in different feature directions, in contrast to isotropic approaches that penalize changes uniformly in all feature directions. By accounting for the empirical second-moment of activations, our approach achieves more precise control while reducing the degradation of model performance on unrelated tasks.

  PublisherDOI

• Training LLM-Based Tutors to Improve Student Learning Outcomes in Dialogues

  Lecture notes in computer science · 2025-01-01 · 11 citations

  book-chapterOpen access
  PublisherOA PDFDOI

• CLEAR-3K: Assessing Causal Explanatory Capabilities in Language Models

  ArXiv.org · 2025-06-20

  preprintOpen access

  We introduce CLEAR-3K, a dataset of 3,000 assertion-reasoning questions designed to evaluate whether language models can determine if one statement causally explains another. Each question present an assertion-reason pair and challenge language models to distinguish between semantic relatedness and genuine causal explanatory relationships. Through comprehensive evaluation of 21 state-of-the-art language models (ranging from 0.5B to 72B parameters), we identify two fundamental findings. First, language models frequently confuse semantic similarity with causality, relying on lexical and semantic overlap instead of inferring actual causal explanatory relationships. Second, as parameter size increases, models tend to shift from being overly skeptical about causal relationships to being excessively permissive in accepting them. Despite this shift, performance measured by the Matthews Correlation Coefficient plateaus at just 0.55, even for the best-performing models.Hence, CLEAR-3K provides a crucial benchmark for developing and evaluating genuine causal reasoning in language models, which is an essential capability for applications that require accurate assessment of causal relationships.

  PublisherOA PDFDOI

• Estimating the Number and Locations of Boundaries in Reverberant Environments with Deep Learning

  2025-03-12

  articleSenior author

  Underwater acoustic environment estimation is a challenging but important task for remote sensing scenarios. Current estimation methods require high signal strength and a solution to the fragile echo labeling problem to be effective. In previous publications, we proposed a general deep learning-based method for two-dimensional environment estimation which outperformed the state-of-the-art, both in simulation and in real-life experimental settings. A limitation of this method was that some prior information had to be provided by the user on the number and locations of the reflective boundaries, and that its neural networks had to be re-trained accordingly for different environments. Utilizing more advanced neural network and time delay estimation techniques, the proposed improved method no longer requires prior knowledge the number of boundaries or their locations, and is able to estimate two-dimensional environments with one or two boundaries. Future work will extend the proposed method to more boundaries and larger-scale environments.

  PublisherDOI

• Turing-Like Test for Personalized Educational AI

  Lecture notes in computer science · 2025-01-01 · 1 citations

  book-chapterSenior author
  PublisherDOI

Recent grants

• Collaborative Research: Design and Analysis of Compressed Sensing DNA Microarrays

  NSF · $316k · 2007–2012

• CIF: Small: Lens-Free Imaging: Can Signal Processing Replace Lenses?

  NSF · $548k · 2015–2018

• Accelerating STEM Learning Through Large-Scale Data Science

  NSF · $5.2M · 2019–2024

• NeTS-NOSS: Adaptivity in Sensor Networks for Optimized Distributed Sensing and Signal Processing

  NSF · $500k · 2005–2011

• Convergence Accelerator Phase I (RAISE): Scalable Knowledge Network to Enable Intelligent Textbooks

  NSF · $1.0M · 2019–2021

Frequent coauthors

• Randall Balestriero

  74 shared

• Andrew Lan

  60 shared

• Rudolf H. Riedi

  HES-SO University of Applied Sciences and Arts Western Switzerland

  52 shared

• Hyeokho Choi

  University of Illinois Urbana-Champaign

  51 shared

• Michael B. Wakin

  47 shared

• Marco F. Duarte

  University of Massachusetts Amherst

  44 shared

• Aswin C. Sankaranarayanan

  Carnegie Mellon University

  42 shared

• Arian Maleki

  42 shared

Labs

• Rice DSP GroupPI

Awards & honors

• Elected Member of the National Academy of Engineering (NAE)…
• Harold W.. McGraw, Jr. Prize in Education (2022)