Upload your resume. PhdFit's six research agents compare your background with faculty profiles, recent publications, lab focus, and outreach opportunities, then rank professors with evidence you can review.
Resume-aware matchingEvidence from recent researchOutreach drafts included
Advisor Match Brief Resume uploaded
Resume signals5 found
NLPHCIFairnessGraph MLPyTorch
Six-agent analysisRunning
Axel · Candidate AnalystResume parsed
Nova · Professor ResearcherRecent papers
Echo · Match ExplainerEvidence linked
Atlas · Strategy AdvisorFit ranked
Lyra · Outreach WriterDraft queued
Juno · Meeting Prep CompilerNext step ready
SC
Dr. Sarah Chen
Stanford · NLP & interpretability
91
fit
Fairness + NLP projects overlap with her recent work.
Strong methods fit: model evaluation and interpretability.
Clear outreach angle for current trustworthy AI research.
Outreach angle
Ask how her lab is extending interpretability methods into fairness audits for real-world AI systems.
Xaq Pitkow
· Electrical and Computer EngineeringVerified
Rice University · Electrical and Computer Engineering
Xaq Pitkow is a faculty member in the Department of Electrical and Computer Engineering at Rice University, specializing in Artificial Intelligence and Machine Learning. His research focuses on advancing understanding and development in these fields, contributing to the academic community through his work at Rice University. He is accessible via email at xaq@rice.edu and is associated with the Rice CS, Rice ECE, and Rice STAT departments, indicating a multidisciplinary approach to his research interests.
To ensure accurate documentation of the implemented models, we clarify several architectural details in the Methods describing the Conv-LSTM and CvT-LSTM architectures.These clarifications are limited to the Methods description and do not affect the results or conclusions.Perspective module: The Methods state that the pupil-position multilayer perceptron MLP uses an 8-dimensional hidden representation; however, in the implemented CvT-LSTM models, this module uses a 16-dimensional hidden representation.
If a person can solve a task, can measuring their brain make it easier to train a model to solve that task too? Recent NeuroAI work suggests that supplementing task training with neural recordings can modestly improve model performance and robustness. However, it is unclear when there should be a benefit from using neural data and how much benefit to expect. We formulate this question mathematically, and begin to address it theoretically using a simple, analytically tractable linear gaussian model of task targets and neural recordings. For a multimodal estimator trained on both brain data and task labels, we derive scaling laws for how performance scales with the numbers of brain and task samples. From these laws we derive relative value and exchange rates between brain samples and task samples, quantifying how much extra task samples neural data is worth as a function of task-brain alignment, neural and task noise, latent dimension, and brain data sample size. We also analyze test distribution shift, to identify conditions where brain-regularized learning can produce substantial robustness gains through learned invariances. Finally, under a fixed collection budget, we characterize the regimes in which brain data is worth collecting. Our results provide a foundation for understanding how valuable brain data could be for improving machine learning.
Representations play a central role in the study of both biological and artificial intelligence, as well as philosophy of mind. Across neuroscience, computer science, and philosophy, a recurring theme is that representations not only carry information but should be ``useful'' for or ``usable'' by an agent in some sense. Here, we review how the ``usefulness'' of representations has been conceptualized and how it figures into different conceptions of representation. We identify and explore four aspects of use and usability: representations generally carry \textit{information}; that information may or may not be \textit{useful} and it may or may not be encoded in a usable \textit{format}; and the representations may or may not be \textit{used downstream}. Building on these four aspects of information and use, we then organize existing perspectives on neural representations into three levels: Representations as Information (Level 1); Representations as Usable (Level 2); and Representations as Used (Level 3). Our account is meant to give readers an appreciation for the diversity of notions of ``neural representation,'' help them navigate the vast and multi-disciplinary literature on the topic, and help them clarify the appropriate notion of representation for their own investigations.
Representations play a central role in the study of both biological and artificial intelligence, as well as philosophy of mind. Across neuroscience, computer science, and philosophy, a recurring theme is that representations not only carry information but should be ``useful'' for or ``usable'' by an agent in some sense. Here, we review how the ``usefulness'' of representations has been conceptualized and how it figures into different conceptions of representation. We identify and explore four aspects of use and usability: representations generally carry \textit{information}; that information may or may not be \textit{useful} and it may or may not be encoded in a usable \textit{format}; and the representations may or may not be \textit{used downstream}. Building on these four aspects of information and use, we then organize existing perspectives on neural representations into three levels: Representations as Information (Level 1); Representations as Usable (Level 2); and Representations as Used (Level 3). Our account is meant to give readers an appreciation for the diversity of notions of ``neural representation,'' help them navigate the vast and multi-disciplinary literature on the topic, and help them clarify the appropriate notion of representation for their own investigations.
If a person can solve a task, can measuring their brain make it easier to train a model to solve that task too? Recent NeuroAI work suggests that supplementing task training with neural recordings can modestly improve model performance and robustness. However, it is unclear when there should be a benefit from using neural data and how much benefit to expect. We formulate this question mathematically, and begin to address it theoretically using a simple, analytically tractable linear gaussian model of task targets and neural recordings. For a multimodal estimator trained on both brain data and task labels, we derive scaling laws for how performance scales with the numbers of brain and task samples. From these laws we derive relative value and exchange rates between brain samples and task samples, quantifying how much extra task samples neural data is worth as a function of task-brain alignment, neural and task noise, latent dimension, and brain data sample size. We also analyze test distribution shift, to identify conditions where brain-regularized learning can produce substantial robustness gains through learned invariances. Finally, under a fixed collection budget, we characterize the regimes in which brain data is worth collecting. Our results provide a foundation for understanding how valuable brain data could be for improving machine learning.
; however, broader connectivity rules remain unknown. Here we leverage the millimetre-scale MICrONS dataset to analyse synaptic connectivity and functional properties of neurons across cortical layers and areas. Our results reveal that neurons with similar response properties are preferentially connected within and across layers and areas-including feedback connections-supporting the universality of 'like-to-like' connectivity across the visual hierarchy. Using a validated digital twin model, we separated neuronal tuning into feature (what neurons respond to) and spatial (receptive field location) components. We found that only the feature component predicts fine-scale synaptic connections beyond what could be explained by the proximity of axons and dendrites. We also discovered a higher-order rule whereby postsynaptic neuron cohorts downstream of presynaptic cells show greater functional similarity than predicted by a pairwise like-to-like rule. Recurrent neural networks trained on a simple classification task develop connectivity patterns that mirror both pairwise and higher-order rules, with magnitudes similar to those in MICrONS data. Ablation studies in these recurrent neural networks reveal that disrupting like-to-like connections impairs performance more than disrupting random connections. These findings suggest that these connectivity principles may have a functional role in sensory processing and learning, highlighting shared principles between biological and artificial systems.
Abstract The complexity of neural circuits makes it challenging to decipher the brain’s algorithms of intelligence. Recent breakthroughs in deep learning have produced models that accurately simulate brain activity, enhancing our understanding of the brain’s computational objectives and neural coding. However, it is difficult for such models to generalize beyond their training distribution, limiting their utility. The emergence of foundation models 1 trained on vast datasets has introduced a new artificial intelligence paradigm with remarkable generalization capabilities. Here we collected large amounts of neural activity from visual cortices of multiple mice and trained a foundation model to accurately predict neuronal responses to arbitrary natural videos. This model generalized to new mice with minimal training and successfully predicted responses across various new stimulus domains, such as coherent motion and noise patterns. Beyond neural response prediction, the model also accurately predicted anatomical cell types, dendritic features and neuronal connectivity within the MICrONS functional connectomics dataset 2 . Our work is a crucial step towards building foundation models of the brain. As neuroscience accumulates larger, multimodal datasets, foundation models will reveal statistical regularities, enable rapid adaptation to new tasks and accelerate research.
bioRxiv (Cold Spring Harbor Laboratory) · 2025-06-03
preprintOpen access
How does the brain process language over time? Research suggests that natural human language is processed hierarchically across brain regions over time. However, attempts to characterize this computation have thus far been limited to tightly controlled experimental settings that capture only a coarse picture of the brain dynamics underlying human natural language comprehension. The recent emergence of LLM encoding models promises a new avenue to discover and characterize rich semantic information in the brain, yet interpretable methods for linking information in LLMs to language processing over time are limited. In this work, we develop a low-rank tensor regression method to decompose LLM encoding models into interpretable components of semantics, time, and brain region activation, and apply the method to a Magnetoencephalography (MEG) dataset in which subjects listened to narrative stories. With only a few components, we show improved performance compared to a standard ridge regression encoding model, suggesting the low-rank models provide a good inductive bias for language encoding. In addition, our method discovers a diverse spectrum of interpretable response components that are sensitive to a rich set of low-level and semantic language features, showing that our method is able to separate distinct language processing features in neural signals. After controlling for low-level audio and sentence features, we demonstrate better capture of semantic features. Through use of low-rank tensor encoding models we are able to decompose neural responses to language features, showing improved encoding performance and interpretable processing components, suggesting our method as a useful tool for uncovering language processes in naturalistic settings.
The growing interest in soft robotics arises from their unique ability to perform tasks beyond the capabilities of rigid robots, with soft actuators playing a central role in this innovation. Among these, electromagnetic soft actuators (ESAs) stand out for their fast response, simple control mechanisms, and compact design. Analytical and experimental studies indicate that smaller ESAs enhance the force per unit cross-sectional area (F/CSA) without compromising force efficiency. This work uses the magnetic vector potential (MVP) to calculate the magnetic field of an ESA, which is then used to derive the actuator’s generated force. A mixed integer non-linear programming (MINLP) optimization framework is introduced to maximize the ESA’s F/CSA. Unlike prior methods that independently optimized parameters, such as ESA length and permanent magnet diameter, this study jointly optimizes these parameters to achieve a more efficient and effective design. To validate the proposed framework, finite element-based COMSOL 5.4 is used to simulate the magnetic field and generated force, ensuring consistency between MVP-based calculations and the physical model. Additionally, simulation results demonstrate the effectiveness of MINLP optimization in identifying the optimal design parameters for maximizing the F/CSA of the ESA. The data and code are available at GitHub Repository.
