University Of California Berkeley

NSF Awards · FY 2025 · 2025-07

Over the past decade, deep learning has evolved from conquering research benchmarks to systems that interact with humans on a daily basis, including in machine translation, healthcare, speech recognition, semi-autonomous vehicles, and automation. However, a large gap exists between its empirical successes and a theoretical understanding of why / when it works. This project aims to close this gap through foundational understanding of deep learning and designing algorithms to improve reliability and data efficiency. More broadly, the societal impact of this project include i) theoretical understanding and design of algorithms relevant to machine learning, ii) education plans that develop a new seminar series and workshops for secondary school teachers, and iii) improving disability accommodation in academia. This project is divided into three different thrusts. The first thrust is to understand the algorithmic regularization effect of algorithms and architectures. Using these insights, the team will design better loss functions and architectures to improve accuracy. The second thrust is to theoretically compare the accuracy of networks trained with stochastic gradient descent against their architecture-induced kernel methods. This comparison may theoretically demonstrate that neural networks can do feature learning, which explains the empirical success of deep learning, and that kernel methods cannot. Finally, the project will study representation learning, and theoretically analyze how deep networks can transfer their representations between different domains. Such a transfer will allow a reduction of the labeled data requirements for deep learning, potentially allowing its application to data-starved domains. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-07

Deep Reinforcement Learning (DRL), which uses neural networks to solve sequential decision-making problems, has made breakthroughs in real-world applications, such as robotics, gaming, healthcare, and transportation systems. However, current theoretical work on reinforcement learning is restricted to problems with a small number of states; as these results do not cover neural networks, they cannot be used to satisfactorily explain the empirical successes of DRL. This project seeks to bridge this gap by building a mathematical foundation for DRL that leverages ideas from approximation theory, control theory, and optimization theory. This will allow the computational and statistical complexity of DRL to be systematically characterized, and will help with designing more efficient and reliable empirical methods. Education and outreach plans are integrated into this project. Specifically, the investigators will mentor graduate and undergraduate students (some through the STARS program for underrepresented groups at the University of washington), develop new courses and monographs, organize research workshops, and develop course materials for a high school data science and artificial intelligence curriculum. This project has three major components. The first thrust identifies which types of guarantees are achievable by policies for different reinforcement learning problem instances. Concretely, this requires investigating how increasingly structured problem instances enable stronger guarantees for policies; this will be done by using, and further developing, tools from non-convex optimization to describe policies that achieve stationary points, local maxima, and global maxima of the reward function. The second thrust takes the perspective of approximation theory and capacity control to investigate how the neural network complexity can be gradually increased to eventually find the most complex sub-family of neural networks that permit sample-efficient algorithms. The third thrust builds upon the knowledge gained in the first two thrusts, and is devoted to the design of computationally efficient algorithms; this will be done by leveraging tools from optimization theory and by making connections with control theory. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2025 · 2025-07

Project summary/abstract The correct implementation of developmental programs depends on information encoded in an organism’s DNA. Mutations in these regions that control genes have been shown to be responsible for ailments such as developmental defects and cancer. While great progress has been made in mapping these regulatory regions and uncovering how their target genes interact with each other, it is still not possible to precisely predict patterns of gene expression in space and time from knowledge of the DNA regulatory sequence of multicellular organisms. The overarching goal of the proposed work is to leverage knowledge about these regulatory regions and the transcription factors that bind to them in order to reach a predictive understanding of the developmental program of the early embryo of the fruit fly Drosophila melanogaster. Armed with recent innovations in (1) theoretical models that predict the mean and variability in gene expression as a function of regulatory sequence and input transcription factor concentration dynamics, and (2) technology to visualize and quantify transcriptional initiation in real time in live, single cells in a fruit fly embryo, the proposed investigations will achieve significant progress toward predictive understanding of transcriptional regulation in development. First, through cycles of experiments and modeling, the proposed studies will uncover how pioneer transcription factors dictates transcriptional onset dynamics by regulating chromatin accessibility to activators and repressors. Second, an experiment–theory discourse that leverages synthetic biology will be used to reach a predictive understanding of how the number, placement and affinity of transcription factor binding sites dictates the rate of transcriptional initiation. Finally, we will focus on single-cell transcriptional dynamics and its characteristic transcriptional bursts in order to shed light on the molecular mechanisms underlying transcription and its control. Specifically, we will use our novel compound-state Hidden Markov model to determine whether Dorsal controls burst size, frequency, amplitude, or some combination thereof, in order to generate hypotheses about the mechanisms of action of this activator and determine whether stable clusters of high Dorsal concentration that we recently discovered play an active role in regulating transcriptional dynamics. These investigations will fuel the theory–experiment dialogue necessary for reaching a predictive understanding of developmental decision-making. I envision that, by revealing the dynamic molecular mechanisms underlying transcriptional control, we will be able to write governing equations for gene regulation and, ultimately, engineer cellular decision-making programs for bioengineering and therapeutic purposes.

NSF Awards · FY 2025 · 2025-06

Understanding human behavior from video is a challenging and transformative area of research, with applications in robotics, assistive technologies, neuroscience, and beyond. Humans do not act in isolation; their actions are shaped by their surroundings, interactions with others, and the objects they use. This project aims to develop a new foundational paradigm for understanding humans in 4D — their 3D state over time — from any type of video. Unlike current methods, this approach integrates people with their physical and social context, enabling a deeper understanding of human activities. By creating a computational framework that can analyze both exo-centric (third-person) and ego-centric (first-person) videos, the project addresses the limitations of existing methods and supports downstream applications such as assistive technologies, wearable AI, and data analysis for neuroscience and practical everyday tasks. The resulting advancements will enable robots to learn from observing humans, assistive technologies to better support users, and wearable devices to provide richer context for human activity, contributing to safer, more effective, and accessible technologies with far-reaching impacts across science, industry, and society. This project will design a scalable, transformer-based model to capture the 4D state of humans and their situational context, including surrounding environments, social interactions, and object use. By leveraging recent advancements in 3D pose estimation, scene reconstruction, and large-scale multimodal models, the research will unify these aspects into a single flexible framework. The approach accommodates various types of video inputs, whether single or multiple views, and will include comprehensive evaluations of its performance. The resulting open-source code, models, and data will provide tools for researchers to advance 4D human understanding and related fields. This project also integrates research with education by developing curricula that combine vision, geometry, and machine learning, and by creating summer research opportunities for a wide range of students, along with accessible online tools to engage a broader audience. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

Combinatorial optimization (CO) problems are pervasive under the hood of modern life. CO problems underlie artificial intelligence, autonomous driving, logistics in healthcare/power grids/transportation, robotic maneuvering, wireless communications, error tolerant data storage, and many other societally important technologies. In recent years, new ways to solve these problems (using "analog oscillator" mechanisms) have emerged that promise far greater solution effectiveness than current techniques can achieve---if appropriate semiconductor "chip" implementations can be devised. The main goal of this project is to design, fabricate and demonstrate such chip implementations, along with systems that utilize them. Achieving this goal will lead to improved efficiencies solving a variety of societally important combinatorial optimization problems. Dissemination and training are also important components of this project. The specific scheme being investigated is called oscillator Ising machines (OIMs). OIM simulations have predicted high success rates solving various combinatorial optimization problems. However, integrated circuit (IC) implementations have had difficulty delivering such predicted levels of performance. In this project, the investigators will identify technological reasons for this discrepancy, and devise measures to address them. A key feature is an IC fabric that supports programmable interconnectivity between analog units. The impact of noise and device variability will be explored, as will specialized IC designs for different types of combinatorial optimization problems. Potential technology and performance benefits offered by novel nanodevices will also be explored. Evaluation metrics will include quality of solution, success rates, and power/energy required. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

Multiphase flows, such as liquid-gas flows, are omnipresent in power plants, pipelines and countless other engineering applications of great societal importance. How these flows interact with surfaces with varying natural or engineered texture is often a determining factor for process reliability and efficiency. However, the results of such multiphase flow and surface interactions are most often described based on fits to data that are only applicable to the specific application over a narrow range of conditions. Therefore, the principal aim of this project is to discover universal dependencies in such complex flows utilizing highly repeatable experiments studying liquid jet impingent on well-characterized textured engineered surfaces. The project also includes significant educational activity by providing undergraduate research opportunities and incorporates scalable outreach that will benefit the scientific and technical education of high school students as well. The engineering community’s fundamental understanding is very limited when it comes to the interaction of multiphase flows with complex surfaces, defined here as surfaces with (spatially or sample-to-sample) varying hydrophobicity and roughness. Such surfaces, when appropriately designed and fabricated, can also retain microscopic pockets of gas (called plastrons) whose state is a function of the flow they are exposed to, and which greatly impact the overall flow dynamics as shown by our preliminary study. This research builds upon the PIs’ prior published work on a complementary study and expands on the preliminary data yielded by the experimental jet impingement method we have developed. Guided by jet impingement and interfacial thermodynamic theories, these methods will be utilized to study boundary layer growth, wetting, and dewetting on random and structured surfaces with varying degrees of hydrophobicity. Elucidating the role of roughness, varying hydrophobicity and state of plastrons on the dynamics of impinging jets, and dewetting in case of gravity vector normal to surface outward, will contribute to the community's understanding of multiphase flow boundary layer flows – in addition to the enhanced understanding of the interaction of impinging jets on complex surfaces, which alone is an important topic of research. The fundamental understanding that the proposed research will contribute to can be readily utilized in numerous applications and have a broad societal impact in the near-term. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2025 · 2025-06

PROJECT SUMMARY Insights into mechanisms of immune regulation have led to breakthroughs in treatments for cancer and autoimmune disease. Regulatory T cells (Tregs), an immunosuppressive subset of CD4+ T cells, are the ultimate immune regulators – acting as gatekeepers of immune responses, critically preventing autoimmune disease, but also hindering our body’s natural immunity toward cancer. While preclinical models have clearly shown that Treg suppression of immune responses against tumor cells can support cancer progression, the role of their counterparts, exTregs, is incompletely understood. ExTregs arise from Tregs but gain pro-inflammatory cytokine production and ultimately lose the expression of the master transcription factor of the Treg lineage, Foxp3. We hypothesize that exTregs are required to promote anti-cancer immunity and autoimmunity when Foxp3+ Tregs are depleted. Therefore, the activation of immune responses when immunosuppression by Tregs is lifted is critically initiated and promoted by exTregs, a novel mechanism to explain autoimmunity and anti-cancer immunity after Treg ablation. In this proposal, we will utilize a genetic Treg lineage tracing system to both monitor exTregs in situ as well as deplete Tregs alone (standard Treg ablation) or use a new genetic model to deplete both exTregs and Tregs simultaneously (DUAL ablation). Hence, we can test whether exTregs are required for tumor control or the onset of autoimmune disease when Tregs are depleted. Ultimately, our findings will be of great importance to understanding the impact of Treg ablation on the immune response and highlight the need for new strategies that target exTregs to ameliorate autoimmune toxicities or augment anti-cancer immune responses.

NIH Research Projects · FY 2026 · 2025-06

PROJECT SUMMARY Immune-based cancer therapies have revolutionized the treatment of cancer. However, immunosuppression within tumors remains a major obstacle for their success. Regulatory T cells (Tregs), an immunosuppressive subset of CD4+ T cells, are heavily recruited to most tumors where they may suppress antitumor immune responses directly within tumors. Tregs may also impose immunosuppression directly in the lymph nodes draining the tumor (tdLN), limiting T cell priming events that generate robust and potent antitumor T cells. Where Tregs function, in the tumor, tdLN, or both, is of critical importance for Treg-targeting cancer therapies because disrupting Tregs systemically will be highly toxic, whereas selectively targeting intratumoral Tregs (IT-Tregs), if effective against cancer, would treat cancers without autoimmune toxicities. Our objective is to decipher whether blocking Treg function in tumors is sufficient for tumor control and to uncover the mechanisms of IT-Treg immunosuppression. We hypothesize that Tregs function directly within tumors to suppress antitumor immune responses by reducing the capacity of dendritic cells (DCs) to endocytose tumor antigen, thereby also reducing T cell priming in the tdLN. Here we propose to investigate this hypothesis by studying experimental and therapeutic methods of IT-Treg ablation, which appear to effectively control primary tumors as well as untreated metastases without inciting autoimmunity. In this proposal we will investigate the mechanism of Treg control of DC antigen uptake (Aim 1), how altered antigen uptake locally in tumors enhances systemic immunity to metastases (Aim 2), and test whether therapeutically translatable Treg-depleting antibodies delivered intratumorally can also increase DC antigen uptake and promote systemic immune control of metastatic cancer (Aim 3). Our findings will uncover new genetic targets to prevent Treg suppression of DC priming of antitumor T cells without autoimmune toxicity.

NIH Research Projects · FY 2026 · 2025-06

Project Summary/Abstract The goal of this proposal is to computationally investigate the role of spatial phase in the selectivity and invariance of responses of primary (V1) and secondary visual cortex (V2) to intermediate-level features in natural images. This modeling-driven approach will further our understanding of phase-related computations in V1 and V2, as well as provide testable predictions to guide experimental design for studies of V1 and V2 physiology. Behaviorally, primates are robustly invariant to transformations of visual objects and scenes [63,64]. For example, given an object under different novel viewing angles, illumination conditions, and positions, humans are gener- ally able to easily recognize an instance of the object. Although the invariant and selective response properties of individual V1 neurons have been relatively well-studied, how these properties are formed between V1 and inferotemporal cortex (IT) is not understood. It has been proposed that the visual system has been adapted to the statistics of the signals in its environment [33, 65]. Therefore, constructing models of the brain that leverage statistical properties of natural images will provide ecologically-driven predictions that can be tested in experiments. Spatial phase and phase structure are perceptually important signals in the visual perception of structure and form [7–12]. To investigate responses to intermediate-level properties in V1 and V2, such as elongated contours, corners, and boundaries, it is essential to approach the problem using a normative model that leverages phase statistics of natural images. V1 complex cells are known to be relatively phase-invariant, as modeled in the standard energy model of com- plex cells [35]. Although the phase-sensitive V1 simple cells have been found to project to V2 [16], how phase information is used in the ventral stream is not clear [17,18]. Constructing a model that explicitly factorizes phase and amplitude responses of units, with respect to natural image stimuli will provide hypotheses for how phase is represented in V1 and V2. This program will develop a first and second layer independently, each optimized to learn the phase and amplitude- based factors that best describe the stimuli. First, the preliminary first layer V1 model will be refined and tested for robustness, then compared to V1 data along with a baseline [22]. Then, after analysis of the joint phase statistics of the V1 unit responses, a second layer will be constructed from those dependencies. The unit invariances and selectivities in the model will provide testable experimental predictions. This modeling work will be performed using the computational resources in Prof. Bruno Olshausen’s theoretical neuroscience group at the University of California, Berkeley. It will be supported by the theoretical neuroscience expertise of Prof. Olshausen, and the physiological and computational expertise of collaborator Prof. Timothy Oleskiw at the University of Regina. If successful, this project could lead to novel computational tools to investigate invariances and selectivity in V1 and V2, which will contribute to our understanding of how the primate brain forms the high-level percepts of shape and form.

NSF Awards · FY 2025 · 2025-06

Principal Investigators (PIs) Stark and Zitrin will identify some of the largest ionized gas bubbles around massive galaxies within the first billion years of cosmic time. The team will conduct a survey using the Binospec spectrograph on the MMT telescope and the Folded-Port Infrared Echellette spectrometer on the Magellan telescope. The PIs will also collaborate with education experts at the Mount Lemmon Sky Center to host professional development workshops for teachers and provide them with human orrery models to use in their classrooms. This project is supported by both the National Science Foundation and the US-Israel Binational Science Foundation to further collaborations between the astrophysical communities in the two countries. PI Stark is a co-PI of the Reionization Era Bright Emission Line Survey (REBELS) using the Atacama Large Millimeter/Submillmeter Array. This project is a complimentary rest-frame ultraviolet spectroscopic survey of the bright REBELS galaxies at z~7. The team will characterize the progress of reionization around these massive galaxies using the brightness of the Lyman-alpha line and fainter diagnostic lines observed in each system. The data will be used to characterize both the nature of the ionized gas in terms of kinematics and chemical enrichment, as well as to determine whether the major source of ionization is an active galactic nucleus or a population of young massive stars. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

This doctoral dissertation project examines the impact of changes in population pressure and dietary diversification on the long-term sustainability of small-scale, self-sufficient societies. The underlying rationale for this project, grounded in empirical research in ethnobotany and ecology, suggests the implementation of multiple dietary plant staples allows food producing societies to increase and stabilize food yields. Diverse food systems can thus constitute a strategic response to episodes of demographic growth. Archaeology can further substantiate these insights because it provides subsistence and population data on the centennial and millennial scales, highlighting successful strategies that have stood the test of time. At present, the link between subsistence diversity and socio-environmental sustainability increasingly draws interdisciplinary attention, largely due to the vulnerabilities of industrial farming. Contemporary farming is predicated on the continued success of high-yielding monocultures, but the biological simplification of food systems has increased dependency on human inputs and susceptibility to change. Understanding the long-term implications of species variability in farming systems is therefore of value in the United States today. The project addresses these issues by looking at two research questions in the context of an archaeological habitation site where both farming and plant foraging took place. The project first determines the extent of demographic change at the site throughout its occupation: this is accomplished through an analysis of wood charcoal remains that have already been excavated. Based on ethnographic observations of fuelwood collection, charcoal remains are expected to reflect population levels because small groups that do not experience resource stress tend to only select a handful of tree species from easy-to-access locations. The investigator therefore measures changes over time in the number of tree species used for fuel, as well as the accessibility of where the fuel was collected. The latter is assessed through oxygen stable isotope analysis, which represents a methodological innovation in archaeobotany. Data concerning the type of fuelwood that was selected over time serves as a proxy for population pressure. A second question focuses on whether the local food system became more diverse to cope with increased population pressure. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2025-06

Project Summary/Abstract Substance use disorders (SUDs) are the most common, costly, and deadly psychiatric conditions1-3. Our approach centers on the development and validation of neuromarkers – measurable, brain-based indicators of pathophysiology, risk, and resilience. In R01DA043690, we used machine learning and fMRI data to create a cross-validated neuromarker for drug craving, a core feature and diagnostic criterion for SUDs in DSM- 51-3 that predicts drug use and relapse4. The resulting Neurobiological Craving Signature (NCS)5 predicts the intensity of cue-induced drug and food craving in new individuals with cocaine, alcohol, and tobacco use disorders (p<0.0002, d=0.93), discriminates between individuals with vs. without SUDs with 82% accuracy5, and is modulated by a regulation strategy drawn from cognitive behavioral treatment (CBT)5. In this proposal, we validate the NCS based on FDA-NIH-defined criteria6 and unpack its underlying neurobiological systems. For this, we have curated a large database including 82 existing cue reactivity fMRI studies (total N=5,475) from our lab, other labs, the ENIGMA consortium, and NCANDA (from the NIMH Data Archive). In Aim 1, we propose to validate the NCS towards research and clinical uses, following the FDA-NIH Biomarker Working Group6 guidelines. We will assess (1a) Generalizability, by testing whether the NCS predicts craving across drugs, stimulus types, sex/gender, and racial categories (N=3,844); (1b) Bias in over/underpredicting selected groups (sex, race, drug type, treatment-seeking status; N=3,844); (1c) Discriminant validity, by comparing the NCS with neuromarkers for other affective states (N=5,475)7-11; (1d) Diagnostic validity, by testing NCS-based discrimination of drug users from non-users (N=1,773); (1e) Predictive validity, by testing if the NCS can predict which individuals will respond to psychological (N=257) or pharmacological treatments (N=562) vs. control; and (1f) Prognostic validity, by testing if the NCS can predict future drug use or relapse after treatment (N=860). To address potential limitations on generalizability as well as any detected bias, we include plans to fine-tune models for subgroups (e.g., sex, race, drug type) if needed. In Aim 2, we will characterize the neurobiology of the NCS and its components, using an interpretable machine learning approach12, 13. We identify key nodes contributing to craving prediction and deploy a series of techniques to identify craving-related brain pathways. We combine multivariate pattern-based pathway identification14 with dynamic connectivity to estimate both population-level and individualized connectomes. We will (2a) test the predictive power of connectomes above and beyond activity patterns; (2b) assess individual variability in pathway strength; (2c) test whether drug cues enhance and increase functional integration; and (2d) identify latent brain states associated with craving and relapse. Deliverables include a refined understanding of connectivity states linked to craving and SUD recovery vs. relapse, and connectivity-based predictive models composed of specific neural pathways that can be validated across species and methods.

NSF Awards · FY 2025 · 2025-06

This REU Site award to The University of California, located in Berkeley, CA, will support the training of 10 students for 10 weeks during the summers of 2025-2027. It is anticipated that a total of 30 students, primarily from schools with limited research opportunities, will be trained in the program. REU scholars will participate in a mentored, hands-on independent research experience in a University of California laboratory conducting experiments on how stress impacts biological systems. Structured curricular activities will provide preparation for graduate training and/or entering the STEM workforce. This project will contribute to developing the biological sciences workforce while building a community of mentors committed to training the next generation of biological scientists. The program will be assessed using surveys developed by the Student Experience in the Research University Consortium and by tracking participant career outcomes. Students should apply to the REU site using NSF ETAP (Education and Training Application: https://etap.nsf.gov). The scientific focus of this REU Site is board and centers on how robustness and resilience shape organism responses to biotic and abiotic stressors. REU scholars will join active research groups in the Departments of Integrative Biology, Molecular Cell Biology, Neurosciences, or Plant and Microbial Biology, working on projects that span distinct levels of biological organization using cutting-edge experimental biology approaches. Potential projects include studying evolutionary genetics in fish, neurobiology of social behaviors in voles, brain plasticity, plant pathogenesis, developmental biology in fruit flies exposed to different stressors, cellular stress responses, and physiology of insect seasonality. For professional development, REU scholars will receive training in scientific communication, oral presentation, writing skills, and responsible conduct of research. Scholars will also participate in career development panel discussions and faculty-led research seminars to understand more about the opportunities in the STEM workforce. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2025-05

PROJECT SUMMARY/ABSTRACT The study embarks on an exploration of the influence of Social Network Cognitive Buffers (SNCBs), comprising both quantifiable and subjective facets of social interaction, on cognitive aging and their potential link to Alzheimer's Disease and Related Dementias (AD/ADRD). Leveraging the unique context of the COVID-19 pandemic, the research aims to assess the cognitive effects of increased social isolation in older adults and the possible recuperative outcomes of restoring social ties post-pandemic. Guided by two primary objectives, the analysis utilizes longitudinal data from the Health and Retirement Study to distinguish between objective social isolation and subjective feelings of loneliness. First, the study examines changes in cognitive aging due to the fluctuation of SNCBs among individuals aged 65 and above, shedding light on the role of social networks in cognitive health. Secondly, it delves into the diverse impacts of reduced social contact and loneliness on cognitive outcomes, probing into the protective potential of varied social connections. In particular, the research focuses on uncovering the mechanisms most responsible for acting as cognitive buffers, exploring the long-term cognitive effects of the pandemic on older adults, and investigating universal mechanisms, such as the value of social network diversity in reducing cognitive decline. By employing causal inference methodologies and panel models, the study aims to pinpoint the specific SNCBs that significantly affect cognitive health. The anticipated findings could highlight the potential benefits of reestablishing social ties in later life stages, contributing not only to the understanding of conditions like AD/ADRD but also to the broader field of aging research. The study's multifaceted approach represents a vital step in advancing the knowledge of social network effects on cognitive health and has the potential to guide public health strategies for older adults. Additionally, it serves as a significant milestone in the development of the candidate, furthering a promising career in aging research.

NIH Research Projects · FY 2026 · 2025-05

Project summary Alzheimer's disease (AD) represents a critical public health burden. Fortunately, recent disease- modifying treatments for AD have demonstrated promising outcomes based on the successful trials of Lecanemab and Donanemab. Nevertheless, how and when these Amyloid-beta- lowering medications most affect cognition is not known. This is clear from the evidence collected from both Lecanemab and Donanemab trials, where not all populations of participants showed significant beneficial effects of anti-Amyloid-beta (Abeta) plaque treatment on cognition. In this context, we aim to use existing datasets to understand AD pathophysiology progression in heterogeneous subpopulations comprehensively. The established insight will reveal optimal treatment windows for various populations, determining the precise timing and target groups for whom reducing Abeta or tau levels is most likely to yield cognitive benefits. In this proposal, we shall harness novel statistical theory and methodology to uncover the heterogeneous causal pathophysiology of Alzheimer's disease with four specific aims in this direction: (1) developing a new statistical method for heterogeneous causal discovery based on the Amyloid cascade hypothesis, (2) developing a new statistical method for data-driven causal discovery to uncover heterogeneous causal relationships among Abeta, tau accumulation, and cognition in various subpopulations, (3) applying the proposed methods in Aim 1-2 across diverse AD datasets and (4) developing software and dissemination research product.

NIH Research Projects · FY 2026 · 2025-05

Project Summary/Abstract Losing the ability to speak and understand language is devastating for patients with aphasia and their families, negatively impacting multiple aspects of life and emotional well-being. Unfortunately, individuals with aphasia rarely regain their language skills in full. Novel approaches are required that boost the effects of traditional therapies, leading to better outcomes for people with aphasia. The proposed study seeks to evaluate the effects of a promising adjuvant intervention – high-intensity physical exercise training - on recovery in aphasia. A new exercise program, specifically designed for individuals with post-stroke motor and language deficits and with documented safety, feasibility and fidelity in this clinical population, will be the means of providing a safe, stroke- and aphasia-friendly high-intensity exercise intervention to achieve optimal physical and behavioral gains. Innovative outcome measures will include not only language and cognitive measures but also measures of motor skills and psychological and psychosocial outcomes that will holistically assess the benefits of physical exercise. Another cutting-edge aspect of the study is the inclusion of advanced physiological fitness measures and neuroimaging metrics of blood flow and cerebrovascular reactivity as outcome measures, to afford a deeper understanding of the exercise-mediated behavioral and neural effects in stroke survivors. We will recruit 110 individuals with aphasia to evaluate the multi-faceted impact of high-intensity exercise on outcomes and compare its effects to an active control condition involving standard-of-care low-intensity exercise. Participants will be randomly assigned to one of the two conditions. The overall format of the control intervention will be very similar to the target intervention, i.e., it will offer the same level of social and activity benefits but without the cardiovascular demands and strengthening components of the high intensity target condition. This will allow us to precisely isolate the key ingredients behind expected exercise-induced changes in language and cognition and definitively test whether the high-intensity aspect of the exercise – the premise of this project – leads to behavioral gains. First, the benefits of high-intensity exercise for language, cognition, motor, and emotional and psychological well-being in individuals with aphasia will be established. Second, it will be determined how exercise-induced changes in physical fitness are related to changes in language and cognitive measures. Third, to advance our knowledge about the brain mechanisms of the observed cognitive and language benefits, exercise-induced neurovascular changes that relate to behavioral improvements will be carefully evaluated. The validated physical exercise intervention resulting from this project will offer a new tool to clinicians seeking to help individuals with aphasia, either as a free-standing program to enhance physical health, cognition, and well-being, or as a supplementary therapy to standard speech-language therapy. Ultimately, this work could significantly alter our thinking about ancillary aphasia therapies that can benefit those with stroke and aphasia through non-traditional means, in this case, a promising, safe, cost-efficient adjunct intervention.

NIH Research Projects · FY 2026 · 2025-05

PROJECT SUMMARY The goal of this program is the discovery of powerfully simplifying synthetic disconnections, strategies, and concepts to prepare biologically important and topologically complex molecules in rapid and diverse fashion. Multiple complex natural products from the plant families Euphorbiaceae and Thymelaeaceae have entered, or successfully completed, human clinical trials as therapeutic agents. Despite their significant documented medicinal relevance toward cancer treatment, HIV eradication, and neurodegenerative disease many complex diterpenes in these families are not available in significant quantities, and unlike many small molecule drug discovery programs, it is difficult to easily mix- and-match structural fragments embedded within these structures. We have developed a new strategy toward natural products from Euphorbiaceae and Thymelaeaceae which is both efficient and convergent. Diverse polycycles of high sp3 content are becoming increasingly important in medicinal chemistry and library screening. We have developed a new strategy to prepare many different polycyclic frameworks from a single starting material using simple reagents under mild conditions. We seek to exploit these finding towards problems in human health. From both of these activities will emerge new therapeutic lead compounds for the treatment of human diseases and new diverse screening fragments that are currently inaccessible. Through this work, trainees will be provided with rigorous and intellectually stimulating training in synthetic organic chemistry and exposure to applications in biomedicine.

NSF Awards · FY 2025 · 2025-05

Quantum light sources will play a fundamental role in the future of technologies including quantum communications, sensing, and computing. Although many types of quantum emitters have been investigated, they all face long standing challenges making it fundamentally difficult to scale quantum systems to larger systems. This work will focus on silicon, an intrinsically scalable material platform and investigate a particular emitter with strong potential for spin-photon interface in the telecommunication band. The multidisciplinary project will contribute to national quantum initiatives and will be coupled with numerous educational objectives including (1) the full participation of underrepresented minorities via presentations at Historically Black Colleges and Universities (HBCUs), and (2) the integration of K-12 students and undergraduate students in the research. Technical description: Solid state quantum emitters require fabrication at the atomic and molecular scales and have suffered from challenges such as reproducibility when fabricated in a host material. Additionally, many quantum sources do not emit in the telecom band and thus require nonlinear processes to make the single photon useful, affecting system energy efficiency. This work will help understand the properties of newly discovered quantum light emitters in silicon. Such emitters can power future quantum networks and computers. The prospect of quantum optics in silicon is an exciting avenue because it has the potential to address the scaling and integration challenges. The exploratory proposal will investigate (1) the manufacturing of the new defect in silicon, (2) the characterization of color centers in silicon, and (3) the spin properties of the defect in silicon. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-05

Both data modeling and agent-based modeling have been found to be promising ways to integrate computing within the science curriculum to explore these types of complex systems, and they are becoming part of standards and curricula in their own right. However, there are still significant challenges in learning the distinct technologies and pedagogical approaches required to successfully integrate these types of modeling into classroom instruction. This project seeks to develop new technological and pedagogical frameworks that can be used to link these two forms of modeling, lowering barriers and offering multiple points for entry into computing for both teachers and students. In this project, middle school students will explore environmental science using data modeling and agent-based modeling. They will also learn about STEM career pathways through sessions with computational environmental scientists. By connecting these practicing scientists with middle school students, the project will introduce computing as relevant and important, model science careers as meaningful, and foster a sense of shared identity for students. Targeting middle school is crucial because this is when students’ career aspirations are shaped, and when their STEM interest and participation drop steeply. The project will design three technology-rich units that integrate data and computational modeling to explore environmental issues using two existing free, web-based software tools that are already used by many middle school students and teachers. The Common Online Data Analysis Platform (CODAP) is a free educational app for data analysis; Modeling + Data (MoDa) is a platform for students to easily program ABMs using domain-specific blocks, and compare them with real-world data for validation. Both tools were developed through prior NSF awards. The developed “Zoom-in on Science sessions” learning program will introduce students to the nature of computational sciences and career pathways in the computational sciences by bringing environmental scientists to classrooms in person or virtually. Using design-based research methodologies, artifact analysis, surveys, interviews, and screen recordings/classroom observations, the project team will study strategies to broaden participation in computing through consequential content and integrated mentorship by addressing the questions i) How do curricular units focused on environmental issues engage students in computing practices to generate scientific explanations? And ii) How do these activities, integrated with Zoom-in-on Science sessions, impact students’ self-efficacy, attributed values, and sense of identity towards science and computing? The project aims to involve 10 in-service teachers and directly impact approximately 500 students from 4 partnering middle schools. This project is funded by the Innovative Technology Experiences for Students and Teachers (ITEST) program, which supports projects that build understandings of practices, program elements, contexts and processes contributing to increasing students' knowledge and interest in science, technology, engineering, and mathematics (STEM) and information and communication technology (ICT) careers. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-05

This EArly-concept Grants for Exploratory Research (EAGER) project will fund research that looks to advance system designs, models, and algorithms for state-aware demand control to significantly enhance shared use of emerging autonomous mobility systems. The rapid expansion of technologies such as robotaxis, robovans, and urban air mobility has the potential to reshape future cities and transform urban transportation. However, this transformation brings significant challenges, particularly managing congestion. Without effective ride-pooling operations, a substantial portion of vehicle miles could involve zero-occupancy trips, an unprecedented phenomenon in transportation networks. On the other hand, autonomous fleets offer unique opportunities through their innovative and flexible vehicle designs, as well as centralized control systems that promote scalable shared-ride experiences. Unlike traditional demand management strategies, this research project seeks to leverage the real-time state of the network to dynamically shape demand, incorporating evolving vehicle and rider trajectories to reflect existing and potential pooling opportunities. These methods have wide-ranging applications for various stakeholders. The methods developed in this project look to help fleet operators create dynamic fare-setting algorithms and technologies that effectively scale ride-pooling operations, and inform local transportation authorities to design dynamic subsidy programs that encourage pooling and alleviate congestion. Beyond passenger transportation and if successful, findings could also be applied to freight bundling and last-mile delivery consolidation, reducing logistics costs and carbon emissions. Additionally, this project will create an educational game focused on shared ride matching and pricing to inspire future generations to address sustainability challenges associated with modern transportation systems. This research project looks to develop novel models and scalable algorithms for stochastic dynamic decision-making in large-scale transportation networks, with a focus on realistic urban-scale settings comprising tens of thousands of nodes and edges that closely reflect the street grids of major metropolitan areas. Methodologically, the project looks to break new ground by integrating Markov decision processes with large-scale networks, enabling them to operate directly on detailed network topologies rather than relying on the coarse, aggregated zones traditionally used in prior research. To tackle the curse of dimensionality issue inherent in city-scale Markov decision processes, the project seeks to develop customized approximation algorithms to enhance scalability. Additionally, it looks to establish novel performance bounds for these approximation policies, providing worst-case guarantees for arbitrary networks and asymptotic guarantees as networks scale from localized, campus-sized areas to entire cities and beyond. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-05

Eran Rabani of the University of California, Berkeley, is supported by an award from the Chemical Theory, Models and Computational Methods program in the Division of Chemistry to develop advanced computational models that predict the optical properties of chiral semiconductor nanocrystals. These materials, which are essential components in next-generation technologies such as filters, sensors, and displays can be engineered to exhibit specific properties by controlling their size, shape, and surface chemistry. Rabani and his research group will develop neural network-based atomistic models to understand how these chiral nanocrystals interact with light, overcoming significant challenges in understanding and predicting chiroptical properties at the nanoscale. This work could lead to breakthroughs in the design of the next generation of optical materials, with tailored chiroptical properties. The research will also provide training opportunities for students in theoretical and computational chemistry, and in nanoscience and nanotechnology, fostering the next generation of scientists and engineers. Eran Rabani will develop advanced neural network-based atomistic models to describe the electronic and optical properties of semiconductor nanocrystals with chiral ligands. These models, validated by experiments, will treat materials and molecules at the same theoretical footing, providing insights into the mechanisms that govern the optical activity of chiral nanocrystals. The research will address computational challenges in predicting chiroptical properties by incorporating the effects of ligand orientation, ordering, and thermal fluctuation on the excitonic dynamical properties. Rabani will also investigate chiral biexcitons and their coupling with chiral microcavities, exploring their potential for applications in quantum optics and photonics. The broader impacts include training students in computational chemistry and nanoscience, as well as engaging in outreach activities to inspire K-12 students in STEM fields. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-04

Bacterial infections remain a leading cause of mortality in both developed and developing nations. Current diagnostic approaches rely on cell culturing to detect and identify bacterial strains, a process that can take days or even weeks, delaying appropriate treatment. Therefore, there is an urgent need for rapid diagnostic methods that eliminate the need for cell culturing. Recent advancements in biotechnology, such as antibody-based bacterial recognition and genetic amplification techniques, have shown high sensitivity and specificity; however, these methods often require costly reagents, specialized equipment, and skilled personnel, making the process labor-intensive, time-consuming, and expensive. This project thus aims to develop a miniaturized, sensitive, and user-friendly sensing platform by leveraging integrated circuit (IC) chips and microfluidic integration. The resulting system will be highly portable, with potential for widespread use in clinics, hospitals, and emergency rooms to enable rapid diagnostics. Additionally, the research incorporates an educational and outreach program to introduce these technologies to students from K-12 through graduate levels, contributing to future engineering workforce in healthcare industry. This project aims to develop electromagnetic (EM) bacteria sensing systems covering RF-to-millimeter-wave frequencies that combine physics-based and biochemistry-based detection methods. The physics-based approach uses broadband dielectric spectroscopy to distinguish bacterial phenotypes by their unique molecular signatures, while the biochemistry-based approach targets strain-specific surface biomarkers tagged with magnetic nanoparticles. Achieving both requires ultrasensitive detection to resolve small spectral variations and to detect single magnetic nanoparticles. The project will achieve these goals with three engineering innovations: (1) Development of ultrasensitive biosensing electronics using high-Q resonators and circuit techniques to overcome traditional noise-power trade-offs, thereby enhancing detection limits. (2) Integration of microfluidic channels within a silicon chip to precisely align the sensing transducers with the underlying electronics to boost the sensitivity. This will be achieved by removing the micro-meter-sized metal routing inside a silicon chip using wet etchants as well as dry etching using ion plasma. (3) System design and integration to perform broadband spectroscopic measurements and magnetic sensing to capture the unique signatures of various bacterial strains and their specific surface biomarkers. The diagnostic accuracy will be further enhanced with deep learning techniques to extract bacteria-specific features. The integration of both platforms will thus enable rapid and accurate bacterial detection for timely treatment. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-04

At least twice during the Cryogenian Period (720-635 million years ago), ice covered the Earth for millions of years in Snowball Earth events. These are the most extreme episodes of climate change in the geological record, but the duration of glaciation and the rate of ice movement remain poorly constrained. The research team will use sediment core of glacial deposits in Antarctica, coupled with drone-based mapping and precise isotopic dating on Snowball Earth deposits in Namibia, to construct a rate-dependent stratigraphic model. They will also develop a research exchange between University of Namibia and UCSB to train students in the field in Namibia and in the lab in California. This training will occur during the course of geological field mapping and sample analysis that is essential context for their research goals of constraining the tempo and nature Snowball Earth, and will prepare students for careers in Geosciences, including critical minerals. The largest uncertainty in the timeline of Cryogenian glaciations is the onset age of the Marinoan glaciation, which is not only important for constraining the rate and nature of glacial processes, but also mechanisms for initiation and deglaciation. Previous studies have suggested that water-lain sediments deposited during Snowball Earth are incompatible with a weak hydrologic cycle. A proposed reconciliation is that during late stages of Snowball Earth with high CO2, ice-sheets could respond to orbital forcing leading to a more active hydrological cycle. However, because these studies have not been able to constrain rate, it is unclear if the apparent cyclicity in these stratigraphic sections is due to Quaternary-like orbital forcing or autogenic processes internal to the ice sheet that have little to do with external forcings. The researcher's preliminary U-Pb zircon geochronology and drone-based mapping data document multiple datable volcanic horizons that can be used to constrain the rate of grounding-line oscillation, and provide the first temporally quantitative facies model for Snowball Earth glacial deposits. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-04

The next communication standard “6G” will make use of new spectrum between 7-24 GHz. However, due to incumbent users, the spectrum will need to be shared in a dynamic way, with channels opening up for short durations, requiring agile radios to hop into and out of bands. This motivates the principal investigator to pursue the design of RF front-end radios that can access very large swaths of bandwidth with programmable center frequency and programmable bandwidth, and tunable in a fast and agile fashion using electronic rather than mechanical tuning. This is in stark contrast to how today’s radios work, as they usually cover much smaller bands (less than 1 GHz) and rely on highly selective filters to minimize interference. The proposed radio would provide wideband tunable functionality by making the radio receiver robust against interference. The transmitter will be equally agile and generate as little out-of-band emissions as possible, so that other users can also share the spectrum. Without using explicit filters, this is a great challenge and requires much innovation in the transmitter architecture. Successful realization of the project would enable resilient broadband connectivity for the foreseeable future, thus broadening wireless access, which is a key national priority. The principal investigator plans on training both graduate students and undergraduates who will participate in the project directly and through classroom interactions. The principal investigator will also continue to teach a chip design “tapeout” course giving undergraduates the opportunity to build radios in advanced CMOS technology nodes and also to test their chips in the lab. This will help with workforce training. The principal investigator is proposing a broadband front-end receiver with an electronically tunable filter that automatically tracks the desired channel of operation using mixer-first N-path techniques, up-converting baseband low-pass filtering to a bandpass response. Linearity is preserved by using voltage feedback in the baseband stages. The principal investigator is also proposing a wideband mixed-signal polar power amplifier that can easily adapt to different bandwidths and modulation schemes due to the flexibility afforded by the digital nature of the architecture. To contend with the high levels of quantization noise and image transmissions, a hybrid PA is proposed that incorporates a low power linear PA that combines with the mixed-signal PA, allowing high efficiency, high linearity, and low spurious transmissions. While N-path filters and digital transmitters have received a lot of interest from the research community, the issue of in-band (rather than out-of-band) linearity of the receiver and out-of-band emissions from a transmitter have been mostly ignored. Also, most of the published work has focused on sub-5 GHz radios. Without addressing these key requirements and new frequency bands, these radio architectures cannot be applied to the envisioned new spectrum ranging from 7-20 GHz. The higher frequency bands necessitate radios without mechanical filters (SAW/BAW/F-BAR) with fast switching between sub-bands. The proposal directly addresses these shortcomings using circuit techniques which allow a new generation of programmable, dynamic, and agile front ends to be used for future spectrum allocations. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-04

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). General audience abstract: Optical atomic clocks are now the most precise and accurate tabletop measurement devices ever constructed by humankind, offering sensitivity to new and exotic physics. The PI has recently developed a new kind of atomic clock apparatus and has used it to demonstrate a comparison between two optical clocks at a precision below one part in 10^19. To give a sense of scale, this corresponds to resolving a difference in the rate the two clocks tick at that would result in them disagreeing with each other by only 1 second after 300 billion years. The PI and a graduate student will use this new apparatus to develop and test ways to use optical atomic clocks to search for dark matter and to detect gravitational waves. This project therefore has the potential to result in new tools for studying the universe through gravitational wave astronomy, and new ways to search for answers to one of the biggest mysteries in physics, the nature of dark matter. The PI will integrate these research topics into new demos and hands-on activities designed to introduce K-12 students to modern physics concepts. Students will engage with these activities at live shows and interactive events as part of the University of Wisconsin “Wonders of Physics” outreach program, with an emphasis on reaching rural communities and Native American reservations in Wisconsin. This project will thereby strengthen public support for modern physics research and help students develop intuition for atomic technologies and their applications. Technical audience abstract: This research project aims to explore and develop emerging applications of optical atomic clocks. The PI has recently demonstrated a first-of-its-kind “multiplexed" optical lattice clock apparatus that enables differential clock comparisons between two or more spatially resolved ensembles of strontium atoms within the same vacuum chamber. These differential measurements eliminate the detrimental effects of clock laser noise and common mode environmental fluctuations, pushing the limits of achievable clock stability and atom-atom coherence. Record differential clock stabilities and fractional frequency precision have now been demonstrated in this apparatus, with a clear path to further gains in performance. The PI and collaborators will use this multiplexed optical lattice clock to develop and demonstrate novel measurement sequences and data analysis techniques for future gravitational wave detection with space-based optical lattice clocks, including the blind injection of simulated gravitational wave signals at realistic strengths. The PI and collaborators will also use the multiplexed optical lattice clock to search for foggy dark matter in previously unexplored regions of parameter space, and to develop new techniques to search for other forms of dark matter. The PI will work with collaborators to develop interactive and engaging demos and inquiry-based activities to introduce K-12 students to modern physics concepts, including the basic principles of atomic clocks and their applications, and will assess their effectiveness using surveys. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.