Search

We apply a variational Ansatz based on neural networks to the problem of spin-$1/2$ fermions in a harmonic trap interacting through a short distance potential. We showed that standard machine learning techniques lead to a quick convergence to the ground state, especially in weakly coupled cases. Higher couplings can be handled efficiently by increasing the strength of interactions during "training"., Comment: 8 pages, 5 figures

We study the poset topology of lattices arising from orientations of 1-skeleta of directionally simple polytopes, with Bruhat interval polytopes $Q_{e,w}$ as our main example. We show that the order complex $\Delta ((u,v)_w)$ of an interval therein is homotopy equivalent to a sphere if $Q_{u,v}$ is a face of $Q_{e,w}$ and is otherwise contractible. This significantly generalizes the known case of the permutahedron. We also show that saturated chains from $u$ to $v$ in such lattices are connected, and in fact highly connected, under moves corresponding to flipping across a 2-face. When $w$ is a Grassmannian permutation, this implies a strengthening of the restriction of Postnikov's move-equivalence theorem to the class of BCFW bridge decomposable plabic graphs., Comment: 14 pages, 1 figure

Diagnostic imaging studies are an increasingly important component of the workup and management of acutely presenting patients. However, ordering appropriate imaging studies according to evidence-based medical guidelines is a challenging task with a high degree of variability between healthcare providers. To address this issue, recent work has investigated if generative AI and large language models can be leveraged to help clinicians order relevant imaging studies for patients. However, it is challenging to ensure that these tools are correctly aligned with medical guidelines, such as the American College of Radiology's Appropriateness Criteria (ACR AC). In this study, we introduce a framework to intelligently leverage language models by recommending imaging studies for patient cases that are aligned with evidence-based guidelines. We make available a novel dataset of patient "one-liner" scenarios to power our experiments, and optimize state-of-the-art language models to achieve an accuracy on par with clinicians in image ordering. Finally, we demonstrate that our language model-based pipeline can be used as intelligent assistants by clinicians to support image ordering workflows and improve the accuracy of imaging study ordering according to the ACR AC. Our work demonstrates and validates a strategy to leverage AI-based software to improve trustworthy clinical decision making in alignment with expert evidence-based guidelines., Comment: 15 pages main text, 4 figures, 1 table

While deep learning has catalyzed breakthroughs across numerous domains, its broader adoption in clinical settings is inhibited by the costly and time-intensive nature of data acquisition and annotation. To further facilitate medical machine learning, we present an ultrasound dataset of 10,223 Brightness-mode (B-mode) images consisting of sagittal slices of porcine spinal cords (N=25) before and after a contusion injury. We additionally benchmark the performance metrics of several state-of-the-art object detection algorithms to localize the site of injury and semantic segmentation models to label the anatomy for comparison and creation of task-specific architectures. Finally, we evaluate the zero-shot generalization capabilities of the segmentation models on human ultrasound spinal cord images to determine whether training on our porcine dataset is sufficient for accurately interpreting human data. Our results show that the YOLOv8 detection model outperforms all evaluated models for injury localization, achieving a mean Average Precision (mAP50-95) score of 0.606. Segmentation metrics indicate that the DeepLabv3 segmentation model achieves the highest accuracy on unseen porcine anatomy, with a Mean Dice score of 0.587, while SAMed achieves the highest Mean Dice score generalizing to human anatomy (0.445). To the best of our knowledge, this is the largest annotated dataset of spinal cord ultrasound images made publicly available to researchers and medical professionals, as well as the first public report of object detection and segmentation architectures to assess anatomical markers in the spinal cord for methodology development and clinical applications.

Sensor fusion combines data from multiple sensor sources to improve reliability, robustness, and accuracy of data interpretation. The Fuzzy Integral (FI), in particular, the Choquet integral (ChI), is often used as a powerful nonlinear aggregator for fusion across multiple sensors. However, existing supervised ChI learning algorithms typically require precise training labels for each input data point, which can be difficult or impossible to obtain. Additionally, prior work on ChI fusion is often based only on the normalized fuzzy measures, which bounds the fuzzy measure values between [0, 1]. This can be limiting in cases where the underlying scales of input data sources are bipolar (i.e., between [-1, 1]). To address these challenges, this paper proposes a novel Choquet integral-based fusion framework, named Bi-MIChI (pronounced "bi-mi-kee"), which uses bi-capacities to represent the interactions between pairs of subsets of the input sensor sources on a bi-polar scale. This allows for extended non-linear interactions between the sensor sources and can lead to interesting fusion results. Bi-MIChI also addresses label uncertainty through Multiple Instance Learning, where training labels are applied to "bags" (sets) of data instead of per-instance. Our proposed Bi-MIChI framework shows effective classification and detection performance on both synthetic and real-world experiments for sensor fusion with label uncertainty. We also provide detailed analyses on the behavior of the fuzzy measures to demonstrate our fusion process., Comment: 10 pages, 7 figures, 7 tables; Accepted to 2024 FUZZ-IEEE and presented at 2024 IEEE WCCI; Code available at https://github.com/hvak/Bi-MIChI

Respiratory diseases impose a significant burden on global health, with current diagnostic and management practices primarily reliant on specialist clinical testing. This work aims to develop machine learning-based algorithms to facilitate at-home respiratory disease monitoring and assessment for patients undergoing continuous positive airway pressure (CPAP) therapy. Data were collected from 30 healthy adults, encompassing respiratory pressure, flow, and dynamic thoraco-abdominal circumferential measurements under three breathing conditions: normal, panting, and deep breathing. Various machine learning models, including the random forest classifier, logistic regression, and support vector machine (SVM), were trained to predict breathing types. The random forest classifier demonstrated the highest accuracy, particularly when incorporating breathing rate as a feature. These findings support the potential of AI-driven respiratory monitoring systems to transition respiratory assessments from clinical settings to home environments, enhancing accessibility and patient autonomy. Future work involves validating these models with larger, more diverse populations and exploring additional machine learning techniques., Comment: 10 pages, 2 figures

This paper proves that the facet-ridge incidence graph of the order complex of any finite geometric lattice of rank $r$ has diameter at most ${r \choose 2}$. A key ingredient is the well-known fact that every ordering of the atoms of any finite geometric lattice gives rise to a lexicographic shelling of its order complex. The paper also gives results that provide some evidence that this bound ought to be sharp as well as examples indicating that the question of sharpness is quite subtle.

Control tuning and adaptation present a significant challenge to the usage of robots in diverse environments. It is often nontrivial to find a single set of control parameters by hand that work well across the broad array of environments and conditions that a robot might encounter. Automated adaptation approaches must utilize prior knowledge about the system while adapting to significant domain shifts to find new control parameters quickly. In this work, we present a general framework for online controller adaptation that deals with these challenges. We combine meta-learning with Bayesian recursive estimation to learn prior predictive models of system performance that quickly adapt to online data, even when there is significant domain shift. These predictive models can be used as cost functions within efficient sampling-based optimization routines to find new control parameters online that maximize system performance. Our framework is powerful and flexible enough to adapt controllers for four diverse systems: a simulated race car, a simulated quadrupedal robot, and a simulated and physical quadrotor. The video and code can be found at https://hersh500.github.io/occam., Comment: 8 pages, 4 figures. Accepted to Conference on Robot Learning (CoRL) 2024

This report summarizes the results of kindergarten through fifth grade students' academic achievement outcomes in mathematics and science following four years of campus-level participation in the Launching Elementary Academic Foundations (LEAF) to STEM program. The LEAF to STEM Program consists of three key components that were adapted for K-5 teachers from a secondary STEM curricular intervention within the same charter system. The three key components were: 1) an elementary-focused project-based learning STEM curriculum, 2) ongoing professional development for participating teachers, and 3) peer-to-peer mentoring with secondary STEM teachers serving as mentors to elementary STEM teachers. The charter school system in which LEAF to STEM was implemented is ethnically and racially diverse, with a higher percentage of students of color than the state in which the system operates (86.1% vs. 71.9% students of color). The system serves a high population of students identified as at-risk, including 60% of students identified as eligible for Free and Reduced Lunch and 24% of students identified as English Language Learners. At the time of randomization, the charter system had a total of 14,572 grades K-5 students on 34 K-5 campuses. The cluster randomized-controlled trial assigned thirty-two K-5 campuses within a large charter system to a treatment/control (business-as-usual) condition using stratified sampling. Campuses were stratified into four groups based on whether a campus's total percentage of students classified as eligible for free-and-reduced price lunch (FRL) fell above or below the system-wide average percentage of FRL students. Within each of those levels, schools were further stratified into two additional groups, based on whether the percentage of a school's students meeting NWEA MAP Reading and Mathematics growth targets in the year prior to randomization to treatment was above or below the system-wide average. Student achievement in mathematics and science was measured using the NWEA MAP Rausch Interval Unit (RIT) scores for mathematics and science for all grades K through five students enrolled in campuses at the time of randomization. Mathematics baseline RIT scores, as well as RIT scores for years one (spring 2020), two (spring 2021), three (spring 2022), and four (spring 2023) of treatment were available for all students, whereas science RIT scores were only available for grades 4 and 5 students after one year (for both 4th and 5th grade students at the onset of the program) or two years of treatment (for 4th grade students only at the onset). Two-level hierarchical linear modeling (students nested in campuses) was employed to explore the impact of treatment on students' mathematics and science achievement (NWEA MAP scores) controlling for demographic background (i.e., ethnicity and gender) and prior achievement at the onset of the LEAF to STEM (beginning of year 1). Results indicated no statistically significant differences in mathematics and science achievement between students in the treatment and control conditions after four years of the exposure to the LEAF to STEM program. When broken down by the grade level, results still did not indicate any significant differences between treatment and control groups in mathematics and science achievement.

We consider the Atiyah-Patodi-Singer (APS) index theorem corresponding to the chiral symmetry of a continuum formulation of staggered fermions called K\"ahler-Dirac fermions, which have been recently investigated as an ingredient in lattice constructions of chiral gauge theories. We point out that there are two notions of chiral symmetry for K\"ahler-Dirac fermions, both having a mixed perturbative anomaly with gravity leading to index theorems on closed manifolds. By formulating these theories on a manifold with boundary, we find the APS index theorems corresponding to each of these symmetries, necessary for a complete picture of anomaly inflow, using a recently discovered physics-motivated proof. We comment on a fundamental difference between the nature of these two symmetries by showing that a sensible local, symmetric boundary condition only exists for one of the two symmetries. This sheds light on how these symmetries behave under lattice discretization, and in particular on their use for recent symmetric-mass generation proposals., Comment: 12 pages, 1 figure

Computer-aided diagnosis systems hold great promise to aid radiologists and clinicians in radiological clinical practice and enhance diagnostic accuracy and efficiency. However, the conventional systems primarily focus on delivering diagnostic results through text report generation or medical image classification, positioning them as standalone decision-makers rather than helpers and ignoring radiologists' expertise. This study introduces an innovative paradigm to create an assistive co-pilot system for empowering radiologists by leveraging Large Language Models (LLMs) and medical image analysis tools. Specifically, we develop a collaborative framework to integrate LLMs and quantitative medical image analysis results generated by foundation models with radiologists in the loop, achieving efficient and safe generation of radiology reports and effective utilization of computational power of AI and the expertise of medical professionals. This approach empowers radiologists to generate more precise and detailed diagnostic reports, enhancing patient outcomes while reducing the burnout of clinicians. Our methodology underscores the potential of AI as a supportive tool in medical diagnostics, promoting a harmonious integration of technology and human expertise to advance the field of radiology.

Multi-modal sensor data fusion takes advantage of complementary or reinforcing information from each sensor and can boost overall performance in applications such as scene classification and target detection. This paper presents a new method for fusing multi-modal and multi-resolution remote sensor data without requiring pixel-level training labels, which can be difficult to obtain. Previously, we developed a Multiple Instance Multi-Resolution Fusion (MIMRF) framework that addresses label uncertainty for fusion, but it can be slow to train due to the large search space for the fuzzy measures used to integrate sensor data sources. We propose a new method based on binary fuzzy measures, which reduces the search space and significantly improves the efficiency of the MIMRF framework. We present experimental results on synthetic data and a real-world remote sensing detection task and show that the proposed MIMRF-BFM algorithm can effectively and efficiently perform multi-resolution fusion given remote sensing data with uncertainty., Comment: 4 pages, 3 figures, 2 tables; Accepted to International Geoscience and Remote Sensing Symposium (IGARSS) 2023; Code available at https://github.com/hvak/MIMRF-BFM

Objective: Information retrieval (IR, also known as search) systems are ubiquitous in modern times. How does the emergence of generative artificial intelligence (AI), based on large language models (LLMs), fit into the IR process? Process: This perspective explores the use of generative AI in the context of the motivations, considerations, and outcomes of the IR process with a focus on the academic use of such systems. Conclusions: There are many information needs, from simple to complex, that motivate use of IR. Users of such systems, particularly academics, have concerns for authoritativeness, timeliness, and contextualization of search. While LLMs may provide functionality that aids the IR process, the continued need for search systems, and research into their improvement, remains essential., Comment: 7 pages, no figures

Self-supervised learning (SSL) methods have shown promise for medical imaging applications by learning meaningful visual representations, even when the amount of labeled data is limited. Here, we extend state-of-the-art contrastive learning SSL methods to 2D+time medical ultrasound video data by introducing a modified encoder and augmentation method capable of learning meaningful spatio-temporal representations, without requiring constraints on the input data. We evaluate our method on the challenging clinical task of identifying lung consolidations (an important pathological feature) in ultrasound videos. Using a multi-center dataset of over 27k lung ultrasound videos acquired from over 500 patients, we show that our method can significantly improve performance on downstream localization and classification of lung consolidation. Comparisons against baseline models trained without SSL show that the proposed methods are particularly advantageous when the size of labeled training data is limited (e.g., as little as 5% of the training set)., Comment: ISBI 2023, 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI)

Multiple data views measured on the same set of participants is becoming more common and has the potential to deepen our understanding of many complex diseases by analyzing these different views simultaneously. Equally important, many of these complex diseases show evidence of subgroup heterogeneity (e.g., by sex or race). HIP (Heterogeneity in Integration and Prediction) is among the first methods proposed to integrate multiple data views while also accounting for subgroup heterogeneity to identify common and subgroup-specific markers of a particular disease. However, HIP is applicable to continuous outcomes and requires programming expertise by the user. Here we propose extensions to HIP that accommodate multi-class, Poisson, and Zero-Inflated Poisson outcomes while retaining the benefits of HIP. Additionally, we introduce an R Shiny application, accessible on shinyapps.io at https://multi-viewlearn.shinyapps.io/HIP_ShinyApp/, that provides an interface with the Python implementation of HIP to allow more researchers to use the method anywhere and on any device. We applied HIP to identify genes and proteins common and specific to males and females that are associated with exacerbation frequency. Although some of the identified genes and proteins show evidence of a relationship with chronic obstructive pulmonary disease (COPD) in existing literature, others may be candidates for future research investigating their relationship with COPD. We demonstrate the use of the Shiny application with a publicly available data. An R-package for HIP would be made available at https://github.com/lasandrall/HIP.

Qubit regularization provides a rich framework to explore quantum field theories. The freedom to choose how the important symmetries of the theory are embedded in the qubit regularization scheme allows us to construct new lattice models with rich phase diagrams. Some of the phases can contain topological terms which lead to critical phases. In this work we introduce and study the SU(3)-F qubit regularization scheme to embed the SO(3) spin-symmetry. We argue that qubit models in this regularization scheme contain several phases including a critical phase which describes the k = 1 Wess-Zumino-Witten (WZW) conformal field theory (CFT) at long distances, and two massive phases one of which is trvially gapped and the other which breaks the lattice translation symmetry. We construct a simple space-time Euclidean lattice model with a single coupling U and study it using the Monte Carlo method. We show the model has a critical phase at small U and a trivially massive phase at large U with a first order transition separating the two. Another feature of our model is that it is symmetric under space-time rotations, which means the temporal and spatial lattice spacing are connected to each other. The unitary time evolution operator obtained by a Wick rotation of the transfer matrix of our model can help us compute the physics of the k = 1 WZW CFT in real time without the need for tuning the temporal lattice spacing to zero. We use this idea to introduce the concept of a relativistic quantum circuit on a discrete space-time lattice., Comment: 15 pages

We give a general derivation of Ginsparg-Wilson relations for both Dirac and Majorana fermions in any dimension. These relations encode continuous and discrete chiral, parity and time reversal anomalies and will apply to the various classes of free fermion topological insulators and superconductors (in the framework of a relativistic quantum field theory in Euclidean spacetime). We show how to formulate the exact symmetries of the lattice action and the relevant index theorems for the anomalies., Comment: 14 pages. Submitted Version. Section III.D revised, and other minor improvements

Neural networks have been proposed as efficient numerical wavefunction ansatze which can be used to variationally search a wide range of functional forms for ground state solutions. These neural network methods are also advantageous in that more variational parameters and system degrees of freedom can be easily added. We benchmark the methodology by using neural networks to study several different integrable bosonic quantum systems in one dimension and compare our results to the exact solutions. While testing the scalability of the procedure to systems with many particles, we also introduce using symmetric function inputs to the neural network to enforce exchange symmetries of indistinguishable particles., Comment: 12 pages, 8 figures

Frame-by-frame annotation of bounding boxes by clinical experts is often required to train fully supervised object detection models on medical video data. We propose a method for improving object detection in medical videos through weak supervision from video-level labels. More concretely, we aggregate individual detection predictions into video-level predictions and extend a teacher-student training strategy to provide additional supervision via a video-level loss. We also introduce improvements to the underlying teacher-student framework, including methods to improve the quality of pseudo-labels based on weak supervision and adaptive schemes to optimize knowledge transfer between the student and teacher networks. We apply this approach to the clinically important task of detecting lung consolidations (seen in respiratory infections such as COVID-19 pneumonia) in medical ultrasound videos. Experiments reveal that our framework improves detection accuracy and robustness compared to baseline semi-supervised models, and improves efficiency in data and annotation usage., Comment: IPMI 2023

We provide quantitative evidence suggesting social learning in sperm whales across socio-cultural boundaries, using acoustic data from the Pacific and Atlantic Oceans. Traditionally, sperm whale populations are categorized into clans based on their vocal repertoire: the rhythmically patterned click sequences (codas) that they use. Among these codas, identity codas function as symbolic markers for each clan, accounting for 35-60% of codas they produce. We introduce a computational method to model whale speech, which encodes rhythmic micro-variations within codas, capturing their vocal style. We find that vocal style-clans closely align with repertoire-clans. However, contrary to vocal repertoire, we show that sympatry increases vocal style similarity between clans for non-identity codas, i.e. most codas, suggesting social learning across cultural boundaries. More broadly, this subcoda structure model offers a framework for comparing communication systems in other species, with potential implications for deeper understanding of vocal and cultural transmission within animal societies.