Search

In physics and engineering, many processes are modeled using non-differentiable black-box simulators, making the optimization of such functions particularly challenging. To address such cases, inspired by the Gradient Theorem, we propose locality-aware surrogate models for active model-based black-box optimization. We first establish a theoretical connection between gradient alignment and the minimization of a Gradient Path Integral Equation (GradPIE) loss, which enforces consistency of the surrogate's gradients in local regions of the design space. Leveraging this theoretical insight, we develop a scalable training algorithm that minimizes the GradPIE loss, enabling both offline and online learning while maintaining computational efficiency. We evaluate our approach on three real-world tasks - spanning automated in silico experiments such as coupled nonlinear oscillators, analog circuits, and optical systems - and demonstrate consistent improvements in optimization efficiency under limited query budgets. Our results offer dependable solutions for both offline and online optimization tasks where reliable gradient estimation is needed.

This study presents a novel approach to human keypoint detection in low-resolution thermal images using transfer learning techniques. We introduce the first application of the Timed Up and Go (TUG) test in thermal image computer vision, establishing a new paradigm for mobility assessment. Our method leverages a MobileNetV3-Small encoder and a ViTPose decoder, trained using a composite loss function that balances latent representation alignment and heatmap accuracy. The model was evaluated using the Object Keypoint Similarity (OKS) metric from the COCO Keypoint Detection Challenge. The proposed model achieves better performance with AP, AP50, and AP75 scores of 0.861, 0.942, and 0.887 respectively, outperforming traditional supervised learning approaches like Mask R-CNN and ViTPose-Base. Moreover, our model demonstrates superior computational efficiency in terms of parameter count and FLOPS. This research lays a solid foundation for future clinical applications of thermal imaging in mobility assessment and rehabilitation monitoring., Comment: Accepted to AICAS 2025. This is the preprint version

Machine learning models are advancing circuit design, particularly in analog circuits. They typically generate netlists that lack human interpretability. This is a problem as human designers heavily rely on the interpretability of circuit diagrams or schematics to intuitively understand, troubleshoot, and develop designs. Hence, to integrate domain knowledge effectively, it is crucial to translate ML-generated netlists into interpretable schematics quickly and accurately. We propose Schemato, a large language model (LLM) for netlist-to-schematic conversion. In particular, we consider our approach in the two settings of converting netlists to .asc files for LTSpice and LATEX files for CircuiTikz schematics. Experiments on our circuit dataset show that Schemato achieves up to 93% compilation success rate for the netlist-to-LaTeX conversion task, surpassing the 26% rate scored by the state-of-the-art LLMs. Furthermore, our experiments show that Schemato generates schematics with a mean structural similarity index measure that is 3xhigher than the best performing LLMs, therefore closer to the reference human design.

Designing integrated circuits involves substantial complexity, posing challenges in revealing its potential applications - from custom digital cells to analog circuits. Despite extensive research over the past decades in building versatile and automated frameworks, there remains open room to explore more computationally efficient AI-based solutions. This paper introduces the graph composer GraCo, a novel method for synthesizing integrated circuits using reinforcement learning (RL). GraCo learns to construct a graph step-by-step, which is then converted into a netlist and simulated with SPICE. We demonstrate that GraCo is highly configurable, enabling the incorporation of prior design knowledge into the framework. We formalize how this prior knowledge can be utilized and, in particular, show that applying consistency checks enhances the efficiency of the sampling process. To evaluate its performance, we compare GraCo to a random baseline, which is known to perform well for smaller design space problems. We demonstrate that GraCo can discover circuits for tasks such as generating standard cells, including the inverter and the two-input NAND (NAND2) gate. Compared to a random baseline, GraCo requires 5x fewer sampling steps to design an inverter and successfully synthesizes a NAND2 gate that is 2.5x faster.

We introduce basic mathematical techniques, followed by an exploration of three distinct topics: the Callan-Giddings-Harvey-Strominger (CGHS) model in 1+1-dimensional spacetime, the formation of astrophysical jets in Schwarzschild-like black holes, and collisions and confinement phenomena in the third-order Lovelock gravity. In the CGHS model, we investigate the collision of ghost fields within the dilaton background geometry, observing the formation and dissolution of wormholes by inserting and removing the ghost fields, respectively. This process mimics a cosmological-scale analogue of Feynman diagrams. Next, we study the non-zero expectation values of bumblebee fields due to Lorentz symmetry breaking. This alteration in the energy-momentum tensor necessitates the inclusion of a potential vacuum, resulting in a shift of the vacuum solution towards Schwarzchild-like black holes with a scaling factor $l$. This scaling factor facilitates discussions on the collision of null sources, leading to the formation of impulsive null shells and satisfying the type-D condition. When $l$ approaches zero, jet-like formations vanish, transforming the problem into one involving colliding gravitational waves, which is isometric to the Schwarzschild geometry. Moreover, our method can be applied to any resembling Schwarzschild-like metrics. We aim to enhance our model by incorporating additional physical factors such as extra polarizations or EM fields. Finally, our examination extends to the 4-dimensional third-order Lovelock gravity, observing that particles possess finite energy and be confined within the metric time interval extending from - to + infinity. Moreover, this finding does not admit flat rotation curves. Additionally, when collisions occur within the background of this metric, intriguingly, we observe impulsive Weyl curvatures along the null boundaries subsequent to the collision.

Interferometric scattering (iSCAT) microscopy is currently among the most powerful techniques available for achieving high-sensitivity single-particle localization. This capability is realized through homodyne detection, where interference with a reference wave offers the promise of exceptionally precise three-dimensional (3D) localization. However, the practical application of iSCAT to 3D tracking has to date been hampered by rapid oscillations in the signal-to-noise ratio (SNR) as particles move along the axial direction. In this study, we introduce a novel strategy based on back pupil plane engineering, wherein we use a spiral phase mask to re-distribute the phase of the scattered field of the particle uniformly across phase space, thus ensuring consistent SNR as the particle moves throughout the focal volume. Our findings demonstrate that this modified spiral phase iSCAT exhibits greatly enhanced localizability characteristics. We substantiate our theoretical results with numerical and experimental demonstrations, showcasing the practical application of this approach for high-precision, ultrahigh-speed (20,000 frames per second) 3D tracking, and characterization of freely diffusing nanoparticles., Comment: 21 pages, 5 figures

The accretion disks around black holes consist of infalling matter boosted almost to the speed of light making collisions with opposite counterpart. This is the rough picture occurring near black holes or other strongly gravitating centers that produce observed phenomena such as astrophysical jets. A toy model that can be considered imitating such a process is colliding null sources in general relativity. We present such a simple model projected into the plane of null coordinates that takes into account only neutral sources. We show that even at such a simplified model, uncharged and non-rotating, it is possible to obtain jet-like ejections albeit they lie below the horizon. In the present study the spacetime consists of either one of i) a cloud of strings, ii) a global monopole, iii) a particular model of bumblebee gravity, all described by a similar class of stress-energy tensor. There are gravitational waves accompanying the null sources and naturally collision of gravitational waves is also taken into account. After the collision, the spacetime contains both null and non-null sources, followed by trailing gravitational radiations. Locally the interaction region of the colliding null-sources and gravitational waves is isometric to the static background spacetime., Comment: Accepted by Physica Scripta

At certain time the Nariai spacetime is split into two parts. The resulting cosmology consists of the original Nariai and a new component with topology $(flat)_2\times S^2$, that is equivalent to a cloud of strings. We explore the properties of our hybrid cosmological model.

Comformally flat spacetime converts the Anti-de Sitter (AdS) spacetime into a z-dependent metric. We take advantage of this metric to solve Dirac equations in the AdS geometry analytically, including Dirac, Majorana, Dirac tachyon and Majorana tachyon. Constraints on Majorana wavefunctions limit the mass. Besides, there gives rise flatlands and Majorana particles hop in between, just mimicking the D-brane stack and open strings. Furthermore, a plausible way of ion-trapped experiment is suggested., Comment: 6 pages, 2 figures

This white paper introduces aoip.ai, a groundbreaking open-source SDK incorporating peer-to-peer technology and advanced AI integration to transform VoIP and IoT applications. It addresses key market challenges by enhancing data security, elevating communication quality, and providing greater flexibility for developers and users. Developed in collaboration with Carnegie Mellon University, aoip.ai sets a new standard for decentralized and democratized communication solutions.

This study introduces Ascle, a pioneering natural language processing (NLP) toolkit designed for medical text generation. Ascle is tailored for biomedical researchers and healthcare professionals with an easy-to-use, all-in-one solution that requires minimal programming expertise. For the first time, Ascle evaluates and provides interfaces for the latest pre-trained language models, encompassing four advanced and challenging generative functions: question-answering, text summarization, text simplification, and machine translation. In addition, Ascle integrates 12 essential NLP functions, along with query and search capabilities for clinical databases. The toolkit, its models, and associated data are publicly available via https://github.com/Yale-LILY/MedGen., Comment: 5 figures, 4 tables

The pion, as the Goldstone boson of the strong interaction, is the lightest QCD bound state and responsible for the long-range nucleon-nucleon interaction inside the nucleus. Our knowledge on the pion partonic structure is limited by the existing Drell-Yan data which are primarily sensitive to the pion valence-quark distributions. The recent progress of global analysis of pion's parton distribution functions (PDFs) utilizing various experimental approaches are introduced. From comparisons between the pion-induced $J/\psi$ and $\psi(2S)$ production data with theoretical calculations using the CEM and NRQCD models, we show how these charmonium production data could provide useful constraints on the pion PDFs., Comment: 17 pages, 6 figures; invited review. arXiv admin note: text overlap with arXiv:2006.06947, arXiv:2209.04072

Falls are the public health issue for the elderly all over the world since the fall-induced injuries are associated with a large amount of healthcare cost. Falls can cause serious injuries, even leading to death if the elderly suffers a "long-lie". Hence, a reliable fall detection (FD) system is required to provide an emergency alarm for first aid. Due to the advances in wearable device technology and artificial intelligence, some fall detection systems have been developed using machine learning and deep learning methods to analyze the signal collected from accelerometer and gyroscopes. In order to achieve better fall detection performance, an ensemble model that combines a coarse-fine convolutional neural network and gated recurrent unit is proposed in this study. The parallel structure design used in this model restores the different grains of spatial characteristics and capture temporal dependencies for feature representation. This study applies the FallAllD public dataset to validate the reliability of the proposed model, which achieves a recall, precision, and F-score of 92.54%, 96.13%, and 94.26%, respectively. The results demonstrate the reliability of the proposed ensemble model in discriminating falls from daily living activities and its superior performance compared to the state-of-the-art convolutional neural network long short-term memory (CNN-LSTM) for FD.

Fall accidents are critical issues in an aging and aged society. Recently, many researchers developed pre-impact fall detection systems using deep learning to support wearable-based fall protection systems for preventing severe injuries. However, most works only employed simple neural network models instead of complex models considering the usability in resource-constrained mobile devices and strict latency requirements. In this work, we propose a novel pre-impact fall detection via CNN-ViT knowledge distillation, namely PreFallKD, to strike a balance between detection performance and computational complexity. The proposed PreFallKD transfers the detection knowledge from the pre-trained teacher model (vision transformer) to the student model (lightweight convolutional neural networks). Additionally, we apply data augmentation techniques to tackle issues of data imbalance. We conduct the experiment on the KFall public dataset and compare PreFallKD with other state-of-the-art models. The experiment results show that PreFallKD could boost the student model during the testing phase and achieves reliable F1-score (92.66%) and lead time (551.3 ms).

In recent years, the demand for pervasive smart services and applications has increased rapidly. Device-free human detection through sensors or cameras has been widely adopted, but it comes with privacy issues as well as misdetection for motionless people. To address these drawbacks, channel state information (CSI) captured from commercialized Wi-Fi devices provides rich signal features for accurate detection. However, existing systems suffer from inaccurate classification under a non-line-of-sight (NLoS) and stationary scenario, such as when a person is standing still in a room corner. In this work, we propose a system called CRONOS (Colorization and Contrastive Learning Enhanced NLoS Human Presence Detection), which generates dynamic recurrence plots (RPs) and color-coded CSI ratios to distinguish mobile and stationary people from vacancy in a room, respectively. We also incorporate supervised contrastive learning to retrieve substantial representations, where consultation loss is formulated to differentiate the representative distances between dynamic and stationary cases. Furthermore, we propose a self-switched static feature enhanced classifier (S3FEC) to determine the utilization of either RPs or color-coded CSI ratios. Our comprehensive experimental results show that CRONOS outperforms existing systems that either apply machine learning or non-learning based methods, as well as non-CSI based features in open literature. CRONOS achieves the highest human presence detection accuracy in vacancy, mobility, line-of-sight (LoS), and NLoS scenarios., Comment: Accepted by IEEE IoT-J

Through the Ernst formalism we provide expression for a class of colliding Einstein-Maxwell (EM) metrics with cross polarization. Local isometry is imposed as a means to transform interaction region of the spacetime into stationary, charged Zipoy-Voorhees (ZV) metric in Schwarzschild coordinates. The ZV-metric is known to describe planetary/stellar objects with arbitrary topology. "The world may be seen in a grain of sand" - William Blake., Comment: A section has been added