Search

Physics-informed neural networks (PINN) is a machine learning (ML)-based method to solve partial differential equations that has gained great popularity due to the fast development of ML libraries in the last few years. The Poisson-Boltzmann equation (PBE) is widely used to model mean-field electrostatics in molecular systems, and in this work we present a detailed investigation of the use of PINN to solve the PBE. Starting from a multidomain PINN for the PBE with an interface, we assess the importance of incorporating different features into the neural network architecture. Our findings indicate that the most accurate architecture utilizes input and output scaling layers, a random Fourier features layer, trainable activation functions, and a loss balancing algorithm. The accuracy of our implementation is of the order of 10$^{-2}$ -- $10^{-3}$, which is similar to previous work using PINN to solve other differential equations. We also explore the possibility of incorporating experimental information into the model, and discuss challenges and future work, especially regarding the nonlinear PBE. Along with this manuscript, we are providing an open-source implementation to easily perform computations from a PDB file. We hope this work will motivate application scientists into using PINN to study molecular electrostatics, as ML technology continues to evolve at a high pace.

Large language models (LLMs) are increasingly employed in information-seeking and decision-making tasks. Despite their broad utility, LLMs tend to generate information that conflicts with real-world facts, and their persuasive style can make these inaccuracies appear confident and convincing. As a result, end-users struggle to consistently align the confidence expressed by LLMs with the accuracy of their predictions, often leading to either blind trust in all outputs or a complete disregard for their reliability. In this work, we explore supervised finetuning on uncertainty-augmented predictions as a method to develop models that produce linguistic expressions of uncertainty. Specifically, we measure the calibration of pre-trained models and then fine-tune language models to generate calibrated linguistic expressions of uncertainty. Through experiments on various question-answering datasets, we demonstrate that LLMs are well-calibrated in assessing their predictions, and supervised finetuning based on the model's own confidence leads to well-calibrated expressions of uncertainty, particularly for single-claim answers.

We propose a flexible and theoretically supported framework for scalable nonnegative matrix factorization. The goal is to find nonnegative low-rank components directly from compressed measurements, accessing the original data only once or twice. We consider compression through randomized sketching methods that can be adapted to the data, or can be oblivious. We formulate optimization problems that only depend on the compressed data, but which can recover a nonnegative factorization which closely approximates the original matrix. The defined problems can be approached with a variety of algorithms, and in particular, we discuss variations of the popular multiplicative updates method for these compressed problems. We demonstrate the success of our approaches empirically and validate their performance in real-world applications.

Low-noise amplifiers are of great importance in the field of quantum technologies. We study a thermally driven parametric amplifier based on a superconductor-insulator-graphene-insulator-superconductor (SIGIS) junction coupled to a superconducting microwave cavity. The strong non-linearity in the temperature dependence of our device leads to thermal self-modulation that produces impedance oscillations at frequencies around twice the angular cavity resonance frequency $\omega_\mathrm{r}$. In particular, reactance modulation of the effective capacitance yields a gain of 18.6 dB over a frequency span of 125 kHz with a minimum noise temperature of $T_\mathrm{N} = 1.4$ K. Our theoretical modelling gives insight into the exact mixing processes, confirmation of the electron-phonon coupling parameter and possible improvements of the studied system., Comment: 9 pages, 5 figures, v2 changes to fig 2 and fig.2 caption

A rapid, exploratory review to establish a shared understanding of what constitutes good quality EdTech and implementation. The review: (1) examined existing frameworks and standards; and (2) identified the characteristics, quality components, essential conditions and evaluation criteria for the design and implementation of EdTech. The report documents: (1) the strength of the evidence base against sector-led standards and frameworks for EdTech product quality; and (2) a single view of the evidence base which could underpin the design of a future framework.

We introduce a family of symmetric convex bodies called generalized ellipsoids of degree $d$ (GE-$d$s), with ellipsoids corresponding to the case of $d=0$. Generalized ellipsoids (GEs) retain many geometric, algebraic, and algorithmic properties of ellipsoids. We show that the conditions that the parameters of a GE must satisfy can be checked in strongly polynomial time, and that one can search for GEs of a given degree by solving a semidefinite program whose size grows only linearly with dimension. We give an example of a GE which does not have a second-order cone representation, but show that every GE has a semidefinite representation whose size depends linearly on both its dimension and degree. In terms of expressiveness, we prove that for any integer $m\geq 2$, every symmetric full-dimensional polytope with $2m$ facets and every intersection of $m$ co-centered ellipsoids can be represented exactly as a GE-$d$ with $d \leq 2m-3$. Using this result, we show that every symmetric convex body can be approximated arbitrarily well by a GE-$d$ and we quantify the quality of the approximation as a function of the degree $d$. Finally, we present applications of GEs to several areas, such as time-varying portfolio optimization, stability analysis of switched linear systems, robust-to-dynamics optimization, and robust polynomial regression.

Background: Incidence of adverse outcome events rises as patients with advanced illness approach end-of-life. Exposures that tend to occur near end-of-life, e.g., use of wheelchair, oxygen therapy and palliative care, may therefore be found associated with the incidence of the adverse outcomes. We propose a strategy for time-to-event analysis to mitigate the time-varying confounding. Methods: We propose a concept of reverse time-to-death (rTTD) and its use for the time-scale in time-to-event analysis. We used data on community-based palliative care uptake (exposure) and emergency department visits (outcome) among patients with advanced cancer in Singapore to illustrate. We compare the results against that of the common practice of using time-on-study (TOS) as time-scale. Results: Graphical analysis demonstrated that cancer patients receiving palliative care had higher rate of emergency department visits than non-recipients mainly because they were closer to end-of-life, and that rTTD analysis made comparison between patients at the same time-to-death. Analysis of emergency department visits in relation to palliative care using TOS time-scale showed significant increase in hazard ratio estimate when observed time-varying covariates were omitted from statistical adjustment (change-in-estimate=0.38; 95% CI 0.15 to 0.60). There was no such change in otherwise the same analysis using rTTD (change-in-estimate=0.04; 95% CI -0.02 to 0.11), demonstrating the ability of rTTD time-scale to mitigate confounding that intensifies in relation to time-to-death. Conclusion: Use of rTTD as time-scale in time-to-event analysis provides a simple and robust approach to control time-varying confounding in studies of advanced illness, even if the confounders are unmeasured., Comment: 22 pages (including 2 tables and 2 figures)

Galvanic coupled-intra-body communication (GC-IBC) is an innovative research area contributing to transform personalized medicine by enabling seamless connectivity and communication among implanted devices. To establish a reliable communication link between implanted devices, the preambles play a crucial role by e.g. conveying syncronization information or supporting channel response estimation. The preambles are carefully designed to ensure that they are mutually orthogonal, to minimize self-interference and maximize separability. For that purpose, many permeable sequences are proposed in the literature for 5G and sensor networks. Golay code, Constant Amplitude Zero Auto Correlation (CAZAC) and Zadoff-Chu (Z-Chu) sequences are among the most popular ones. In this work, we performed a comparative analysis of these sequences to determine their suitability for the GC-IBC system. We evaluated the effectiveness of the preamble sequences on the basis of their correlation properties and probability of error., Comment: Submitted to the 11th ACM International Conference on Nanoscale Computing and Communication (ACM NanoCom 2024), Milan, Italy, October 28-30, 2024

Recently, outstanding identification rates in image classification tasks were achieved by convolutional neural networks (CNNs). to use such skills, selective CNNs trained on a dataset of well-known images of metal surface defects captured with an RGB camera. Defects must be detected early to take timely corrective action due to production concerns. For image classification up till now, a model-based method has been utilized, which indicated the predicted reflection characteristics of surface defects in comparison to flaw-free surfaces. The problem of detecting steel surface defects has grown in importance as a result of the vast range of steel applications in end-product sectors such as automobiles, households, construction, etc. The manual processes for detections are time-consuming, labor-intensive, and expensive. Different strategies have been used to automate manual processes, but CNN models have proven to be the most effective rather than image processing and machine learning techniques. By using different CNN models with fine-tuning, easily compare their performance and select the best-performing model for the same kinds of tasks. However, it is important that using different CNN models either from fine tuning can be computationally expensive and time-consuming. Therefore, our study helps the upcoming researchers to choose the CNN without considering the issues of model complexity, performance, and computational resources. In this article, the performance of various CNN models with transfer learning techniques are evaluated. These models were chosen based on their popularity and impact in the field of computer vision research, as well as their performance on benchmark datasets. According to the outcomes, DenseNet201 outperformed the other CNN models and had the greatest detection rate on the NEU dataset, falling in at 98.37 percent.

Low Earth orbit (LEO) satellite mega constellations are beginning to include laser inter-satellite links (LISLs) to extend the Internet to the most remote locations on Earth. Since the process of establishing these links incurs a setup delay on the order of seconds, a static network topology is generally established well in advance, which is then used for the routing calculations. However, this involves keeping links active even when they are not being used to forward traffic, leading to poor energy efficiency. Motivated by technological advances that are gradually decreasing the LISL setup delays, we foresee scenarios where it will be possible to compute routes and establish dynamic LISLs on demand. This will require considering setup delays as penalties that will affect the end-to-end latency. In this paper, we present a nonlinear optimization model that considers these penalties in the cost function and propose three heuristic algorithms that solve the problem in a tractable way. The algorithms establish different trade-offs in terms of performance and computational complexity. We extensively analyze metrics including average latency, route change rate, outage probability, and jitter in Starlink's Phase I version 2 constellation. The results show the benefit of adaptive routing schemes according to the link setup delay. In particular, more complex schemes can decrease the average end-to-end latency in exchange for an increase in execution time. On the other hand, depending on the maximum tolerated latency, it is possible to use less computationally complex schemes which will be more scalable for the satellite mega constellations of the future.

In this work, we analyze various scaling limits of the training dynamics of transformer models in the feature learning regime. We identify the set of parameterizations that admit well-defined infinite width and depth limits, allowing the attention layers to update throughout training--a relevant notion of feature learning in these models. We then use tools from dynamical mean field theory (DMFT) to analyze various infinite limits (infinite key/query dimension, infinite heads, and infinite depth) which have different statistical descriptions depending on which infinite limit is taken and how attention layers are scaled. We provide numerical evidence of convergence to the limits and discuss how the parameterization qualitatively influences learned features., Comment: Updating for Neurips 2024

A major challenge facing the world is the provision of equitable and universal access to quality education. Recent advances in generative AI (gen AI) have created excitement about the potential of new technologies to offer a personal tutor for every learner and a teaching assistant for every teacher. The full extent of this dream, however, has not yet materialised. We argue that this is primarily due to the difficulties with verbalising pedagogical intuitions into gen AI prompts and the lack of good evaluation practices, reinforced by the challenges in defining excellent pedagogy. Here we present our work collaborating with learners and educators to translate high level principles from learning science into a pragmatic set of seven diverse educational benchmarks, spanning quantitative, qualitative, automatic and human evaluations; and to develop a new set of fine-tuning datasets to improve the pedagogical capabilities of Gemini, introducing LearnLM-Tutor. Our evaluations show that LearnLM-Tutor is consistently preferred over a prompt tuned Gemini by educators and learners on a number of pedagogical dimensions. We hope that this work can serve as a first step towards developing a comprehensive educational evaluation framework, and that this can enable rapid progress within the AI and EdTech communities towards maximising the positive impact of gen AI in education.

Large language models are effective at few-shot in-context learning (ICL). Recent advancements in multimodal foundation models have enabled unprecedentedly long context windows, presenting an opportunity to explore their capability to perform ICL with many more demonstrating examples. In this work, we evaluate the performance of multimodal foundation models scaling from few-shot to many-shot ICL. We benchmark GPT-4o and Gemini 1.5 Pro across 14 datasets spanning multiple domains (natural imagery, medical imagery, remote sensing, and molecular imagery) and tasks (image classification, visual QA, and object localization). We observe that many-shot ICL, including up to almost 2,000 demonstrating examples, leads to substantial improvements compared to few-shot (<100 examples) ICL across all of the datasets. Further, Gemini 1.5 Pro performance continues to improve log-linearly up to the maximum number of tested examples on many datasets. We also find open-weights multimodal foundation models like Llama 3.2-Vision do not benefit from the demonstrating examples, highlighting an important gap between open and closed multimodal foundation models. Given the high inference costs required for many-shot ICL, we also explore the impact of batching multiple queries in a single API call. We show that batching up to 50 queries can lead to performance improvements under zero-shot and many-shot ICL, with substantial gains in the zero-shot setting on multiple datasets, while drastically reducing per-query cost and latency. Finally, while GPT-4o and Gemini 1.5 Pro achieve similar zero-shot performance across the datasets, Gemini 1.5 Pro learns more quickly than GPT-4o on most datasets. Our results suggest that many-shot ICL could enable users to efficiently adapt multimodal foundation models to new applications and domains. Our codebase is publicly available at https://github.com/stanfordmlgroup/ManyICL .

Photonics is a promising technology to accelerate Deep Neural Networks as it can use optical interconnects to reduce data movement energy and it enables low-energy, high-throughput optical-analog computations. To realize these benefits in a full system (accelerator + DRAM), designers must ensure that the benefits of using the electrical, optical, analog, and digital domains exceed the costs of converting data between domains. Designers must also consider system-level energy costs such as data fetch from DRAM. Converting data and accessing DRAM can consume significant energy, so to evaluate and explore the photonic system space, there is a need for a tool that can model these full-system considerations. In this work, we show that similarities between Compute-in-Memory (CiM) and photonics let us use CiM system modeling tools to accurately model photonics systems. Bringing modeling tools to photonics enables evaluation of photonic research in a full-system context, rapid design space exploration, co-design, and comparison between systems. Using our open-source model, we show that cross-domain conversion and DRAM can consume a significant portion of photonic system energy. We then demonstrate optimizations that reduce conversions and DRAM accesses to improve photonic system energy efficiency by up to 3x., Comment: Published at ISPASS 2024

We present an analysis of high angular resolution images of the microlensing target MOA-2007-BLG-192 using Keck adaptive optics and the Hubble Space Telescope. The planetary host star is robustly detected as it separates from the background source star in nearly all of the Keck and Hubble data. The amplitude and direction of the lens-source separation allows us to break a degeneracy related to the microlensing parallax and source radius crossing time. Thus, we are able to reduce the number of possible solutions by a factor of ${\sim}2$, demonstrating the power of high angular resolution follow-up imaging for events with sparse light curve coverage. Following Bennett et al. 2023, we apply constraints from the high resolution imaging on the light curve modeling to find host star and planet masses of $M_{\textrm{host}} = 0.28 \pm 0.04M_{\odot}$ and $m_p = 12.49^{+65.47}_{-8.03}M_{\oplus}$ at a distance from Earth of $D_L = 2.16 \pm 0.30\,$kpc. This work illustrates the necessity for the Nancy Grace Roman Galactic Exoplanet Survey (RGES) to use its own high resolution imaging to inform light curve modeling for microlensing planets that the mission discovers., Comment: 25 pages, 12 figures, accepted in AJ

This paper explores the evolving relationship between clinician trust in LLMs, the transformation of data sources from predominantly human-generated to AI-generated content, and the subsequent impact on the precision of LLMs and clinician competence. One of the primary concerns identified is the potential feedback loop that arises as LLMs become more reliant on their outputs for learning, which may lead to a degradation in output quality and a reduction in clinician skills due to decreased engagement with fundamental diagnostic processes. While theoretical at this stage, this feedback loop poses a significant challenge as the integration of LLMs in healthcare deepens, emphasizing the need for proactive dialogue and strategic measures to ensure the safe and effective use of LLM technology. A key takeaway from our investigation is the critical role of user expertise and the necessity for a discerning approach to trusting and validating LLM outputs. The paper highlights how expert users, particularly clinicians, can leverage LLMs to enhance productivity by offloading routine tasks while maintaining a critical oversight to identify and correct potential inaccuracies in AI-generated content. This balance of trust and skepticism is vital for ensuring that LLMs augment rather than undermine the quality of patient care. Moreover, we delve into the potential risks associated with LLMs' self-referential learning loops and the deskilling of healthcare professionals. The risk of LLMs operating within an echo chamber, where AI-generated content feeds into the learning algorithms, threatens the diversity and quality of the data pool, potentially entrenching biases and reducing the efficacy of LLMs., Comment: 1 figure

This report explores the use of kernel-bypass networking in FaaS runtimes and demonstrates how using Junction, a novel kernel-bypass system, as the backend for executing components in faasd can enhance performance and isolation. Junction achieves this by reducing network and compute overheads and minimizing interactions with the host operating system. Junctiond, the integration of Junction with faasd, reduces median and P99 latency by 37.33% and 63.42%, respectively, and can handle 10 times more throughput while decreasing latency by 2x at the median and 3.5 times at the tail.

The order of training samples can have a significant impact on the performance of a classifier. Curriculum learning is a method of ordering training samples from easy to hard. This paper proposes the novel idea of a curriculum learning approach called Data Distribution-based Curriculum Learning (DDCL). DDCL uses the data distribution of a dataset to build a curriculum based on the order of samples. Two types of scoring methods known as DDCL (Density) and DDCL (Point) are used to score training samples thus determining their training order. DDCL (Density) uses the sample density to assign scores while DDCL (Point) utilises the Euclidean distance for scoring. We evaluate the proposed DDCL approach by conducting experiments on multiple datasets using a neural network, support vector machine and random forest classifier. Evaluation results show that the application of DDCL improves the average classification accuracy for all datasets compared to standard evaluation without any curriculum. Moreover, analysis of the error losses for a single training epoch reveals that convergence is faster when using DDCL over the no curriculum method.

The entanglement dynamics of an exactly solvable, pure dephasing model are studied. Repeated projective measurements are performed on the two-qubit system. Due to the system-environment interaction, system-environment correlations are established between each measurement. Consequently, the environment state keeps evolving. We investigate the effect of this changing environment state on the entanglement dynamics. In particular, we compare the dynamics with the case where the environment state is repeatedly reset., Comment: 12 pages. Comments and remarks most welcome

Clouds play a significant role in global temperature regulation through their effect on planetary albedo. Anthropogenic emissions of aerosols can alter the albedo of clouds, but the extent of this effect, and its consequent impact on temperature change, remains uncertain. Human-induced clouds caused by ship aerosol emissions, commonly referred to as ship tracks, provide visible manifestations of this effect distinct from adjacent cloud regions and therefore serve as a useful sandbox to study human-induced clouds. However, the lack of large-scale ship track data makes it difficult to deduce their general effects on cloud formation. Towards developing automated approaches to localize ship tracks at scale, we present CloudTracks, a dataset containing 3,560 satellite images labeled with more than 12,000 ship track instance annotations. We train semantic segmentation and instance segmentation model baselines on our dataset and find that our best model substantially outperforms previous state-of-the-art for ship track localization (61.29 vs. 48.65 IoU). We also find that the best instance segmentation model is able to identify the number of ship tracks in each image more accurately than the previous state-of-the-art (1.64 vs. 4.99 MAE). However, we identify cases where the best model struggles to accurately localize and count ship tracks, so we believe CloudTracks will stimulate novel machine learning approaches to better detect elongated and overlapping features in satellite images. We release our dataset openly at {zenodo.org/records/10042922}., Comment: 11 pages, 5 figures, submitted to Journal of Machine Learning Research

In free-space optical satellite networks (FSOSNs), satellites connected via laser inter-satellite links (LISLs), latency is a critical factor, especially for long-distance inter-continental connections. Since satellites depend on solar panels for power supply, power consumption is also a vital factor. We investigate the minimization of total network latency (i.e., the sum of the network latencies of all inter-continental connections in a time slot) in a realistic model of a FSOSN, the latest version of the Starlink Phase 1 Version 3 constellation. We develop mathematical formulations of the total network latency over different LISL ranges and different satellite transmission power constraints for multiple simultaneous inter-continental connections. We use practical system models for calculating network latency and satellite optical link transmission power, and we formulate the problem as a binary integer linear program. The results reveal that, for satellite transmission power limits set at 0.5 W, 0.3 W, and 0.1 W, the average total network latency for all five inter-continental connections studied in this work levels off at 339 ms, 361 ms, and 542 ms, respectively. Furthermore, the corresponding LISL ranges required to achieve these average total network latency values are 4500 km, 3000 km, and 1731 km, respectively. Different limitations on satellite transmission power exhibit varying effects on average total network latency (over 100 time slots), and they also induce differing changes in the corresponding LISL ranges. In the absence of satellite transmission power constraints, as the LISL range extends from the minimum feasible range of 1575 km to the maximum feasible range of 5016 km, the average total network latency decreases from 589 ms to 311 ms., Comment: Accepted for publication in IEEE Open Journal of the Communications Society

In free-space optical satellite networks (FSOSNs), satellites can have different laser inter-satellite link (LISL) ranges for connectivity. Greater LISL ranges can reduce network latency of the path but can also result in an increase in transmission power for satellites on the path. Consequently, this tradeoff between satellite transmission power and network latency should be investigated, and in this work we examine it in FSOSNs drawing on the Starlink Phase 1 Version 3 and Kuiper Shell 2 constellations for different LISL ranges and different inter-continental connections. We use appropriate system models for calculating the average satellite transmission power and network latency. The results show that the mean network latency decreases and mean average satellite transmission power increases with an increase in LISL range. For the Toronto--Sydney inter-continental connection in an FSOSN with Starlink's Phase 1 Version 3 constellation, when the LISL range is approximately 2,900 km, the mean network latency and mean average satellite transmission power intersect are approximately 135 ms and 380 mW, respectively. For an FSOSN with the Kuiper Shell 2 constellation in this inter-continental connection, this LISL range is around 3,800 km, and the two parameters are approximately 120 ms and 700 mW, respectively. For the Toronto--Istanbul and Toronto--London inter-continental connections, the LISL ranges at the intersection are different and vary from 2,600 km to 3,400 km. Furthermore, we analyze outage probability performance of optical uplink/downlink due to atmosphere attenuation and turbulence., Comment: Accepted for publication in IEEE Open Journal of Vehicular Technology

We present generalizations of Newton's method that incorporate derivatives of an arbitrary order $d$ but maintain a polynomial dependence on dimension in their cost per iteration. At each step, our $d^{\text{th}}$-order method uses semidefinite programming to construct and minimize a sum of squares-convex approximation to the $d^{\text{th}}$-order Taylor expansion of the function we wish to minimize. We prove that our $d^{\text{th}}$-order method has local convergence of order $d$. This results in lower oracle complexity compared to the classical Newton method. We show on numerical examples that basins of attraction around local minima can get larger as $d$ increases. Under additional assumptions, we present a modified algorithm, again with polynomial cost per iteration, which is globally convergent and has local convergence of order $d$.