Search

Despite the rising prevalence of deep neural networks (DNNs) in cyber-physical systems, their vulnerability to adversarial bit-flip attacks (BFAs) is a noteworthy concern. This paper proposes B3FA, a semi-black-box BFA-based parameter attack on DNNs, assuming the adversary has limited knowledge about the model. We consider practical scenarios often feature a more restricted threat model for real-world systems, contrasting with the typical BFA models that presuppose the adversary's full access to a network's inputs and parameters. The introduced bit-flip approach utilizes a magnitude-based ranking method and a statistical re-construction technique to identify the vulnerable bits. We demonstrate the effectiveness of B3FA on several DNN models in a semi-black-box setting. For example, B3FA could drop the accuracy of a MobileNetV2 from 69.84% to 9% with only 20 bit-flips in a real-world setting.

Formal Property Verification (FPV), using SystemVerilog Assertions (SVA), is crucial for ensuring the completeness of design with respect to the specification. However, writing SVA is a laborious task and has a steep learning curve. In this work, we present a large language model (LLM) -based flow to automatically generate high-quality SVA from the design specification documents, named \ToolName. We introduce a novel sub-task-focused fine-tuning approach that effectively addresses functionally incorrect assertions produced by baseline LLMs, leading to a remarkable 7.3-fold increase in the number of functionally correct assertions. Recognizing the prevalence of syntax and semantic errors, we also developed an iterative refinement method that enhances the LLM's initial outputs by systematically re-prompting it to correct identified issues. This process is further strengthened by a custom compiler that generates meaningful error messages, guiding the LLM towards improved accuracy. The experiments demonstrate a 26\% increase in the number of assertions free from syntax errors using this approach, showcasing its potential to streamline the FPV process.

TinyML models often operate in remote, dynamic environments without cloud connectivity, making them prone to failures. Ensuring reliability in such scenarios requires not only detecting model failures but also identifying their root causes. However, transient failures, privacy concerns, and the safety-critical nature of many applications-where systems cannot be interrupted for debugging-complicate the use of raw sensor data for offline analysis. We propose DEBUG-HD, a novel, resource-efficient on-device debugging approach optimized for KB-sized tinyML devices that utilizes hyper-dimensional computing (HDC). Our method introduces a new HDC encoding technique that leverages conventional neural networks, allowing DEBUG-HD to outperform prior binary HDC methods by 27% on average in detecting input corruptions across various image and audio datasets., Comment: Accepted at the Machine Learning for Systems Workshop at NeurIPS 2024

Direct air capture (DAC) of carbon dioxide is a critical technology for mitigating climate change, but current materials face limitations in efficiency and scalability. We discover novel DAC materials using a combined machine learning (ML) and high-throughput atomistic modeling approach. Our ML model accurately predicts high-quality, density functional theory-computed CO$_{2}$ binding enthalpies for a wide range of nitrogen-bearing moieties. Leveraging this model, we rapidly screen over 1.6 million binding sites from a comprehensive database of theoretically feasible molecules to identify materials with superior CO$_{2}$ binding properties. Additionally, we assess the synthesizability and experimental feasibility of these structures using established ML metrics, discovering nearly 2,500 novel materials suitable for integration into DAC devices. Altogether, our high-fidelity database and ML framework represent a significant advancement in the rational development of scalable, cost-effective carbon dioxide capture technologies, offering a promising pathway to meet key targets in the global initiative to combat climate change., Comment: 18 pages, 6 figures

Assuming that every hyperbolic group is residually finite, we prove the congruence subgroup property for mapping class groups of hyperbolic surfaces of finite type. Under the same assumption, it follows that profinitely equivalent hyperbolic 3-manifolds are commensurable., Comment: 22 pages. Primary article by Wilton, with an appendix by Sisto. v2 incorporates comments on the first version. It includes a more refined profinite-rigidity statement (Theorem B(ii)), proved in the new section 5

Giardia lamblia is a flagellate intestinal protozoan with global distribution causing the disease known as giardiasis. This parasite is responsable for 35.1% of outbreaks of diarrhea caused by contaminated water which and mainly affects children in whom it can cause physical and cognitive impairment. In this paper, we consider a model of population dynamics to represent the behavior of Giardia lamblia in vitro, taking into account its mutation characteristic that guarantees to the protozoan resistance to the drug metronidazole. Different from what is found in the literature, it is pursued as the control objective the extermination of the protozoan considering that the parameters of the model are uncertain and only the partial measurement of the state vector is possible. On these assumptions, a control law is designed and the stability of the closed-loop system is rigorously proved. Simulation and experimental results illustrate the benefits of the proposed population control method of Giardia lamblia., Comment: 7 pages, 5 figures, 1 Table

Low-precision weights and activations in deep neural networks (DNNs) outperform their full-precision counterparts in terms of hardware efficiency. When implemented with low-precision operations, specifically in the extreme case where network parameters are binarized (i.e. BNNs), the two most frequently mentioned benefits of quantization are reduced memory consumption and a faster inference process. In this paper, we introduce a third advantage of very low-precision neural networks: improved fault-tolerance attribute. We investigate the impact of memory faults on state-of-the-art binary neural networks (BNNs) through comprehensive analysis. Despite the inclusion of floating-point parameters in BNN architectures to improve accuracy, our findings reveal that BNNs are highly sensitive to deviations in these parameters caused by memory faults. In light of this crucial finding, we propose a technique to improve BNN dependability by restricting the range of float parameters through a novel deliberately uniform quantization. The introduced quantization technique results in a reduction in the proportion of floating-point parameters utilized in the BNN, without incurring any additional computational overheads during the inference stage. The extensive experimental fault simulation on the proposed BNN architecture (i.e. ZOBNN) reveal a remarkable 5X enhancement in robustness compared to conventional floating-point DNN. Notably, this improvement is achieved without incurring any computational overhead. Crucially, this enhancement comes without computational overhead. \ToolName~excels in critical edge applications characterized by limited computational resources, prioritizing both dependability and real-time performance.

Let $X$ be a non-positively curved cube complex with hyperbolic fundamental group. We prove that $\pi_1(X)$ has a non-free subgroup of infinite index unless $\pi_1(X)$ is either free or a surface group, answering a question of Wise. A similar result for one-relator groups follows, answering a question posed by several authors. The proof relies on a careful analysis of free and cyclic splittings of cubulated groups., Comment: 44 pages

In addition to a plethora of emergent phenomena, the spatial topology of optical vortices enables an array of applications spanning communications to quantum photonics. Nonlinear optics is essential in this context, providing access to an infinitely large set of quantum states associated with the orbital angular momentum of light. Nevertheless, the realization of such processes have failed to keep pace with the ever-growing need to shrink the fundamental length-scale of photonic technologies to the nanometer regime6. Here, we push the boundaries of vortex nonlinear optics to the ultimate limits of material dimensionality. By exploiting second and third-order frequency-mixing processes in semiconducting monolayers, we demonstrate the independent manipulation of the wavelength, orbital angular momentum, and spatial distribution of vortex light-fields. Due to the atomically-thin nature of the host quantum material, this control spans a broad spectral bandwidth in a highly-integrable platform, unconstrained by the traditional limits of bulk nonlinear optical materials. Our work heralds a new avenue for ultra-compact and scalable hybrid nanotechnologies empowered by twisted nonlinear light-matter interactions in van der Waals quantum nanomaterials.

Uncertainty quantification (UQ) provides a resource-efficient solution for on-device monitoring of tinyML models deployed without access to true labels. However, existing UQ methods impose significant memory and compute demands, making them impractical for ultra-low-power, KB-sized TinyML devices. Prior work has attempted to reduce overhead by using early-exit ensembles to quantify uncertainty in a single forward pass, but these approaches still carry prohibitive costs. To address this, we propose QUTE, a novel resource-efficient early-exit-assisted ensemble architecture optimized for tinyML models. QUTE introduces additional output blocks at the final exit of the base network, distilling early-exit knowledge into these blocks to form a diverse yet lightweight ensemble. We show that QUTE delivers superior uncertainty quality on tiny models, achieving comparable performance on larger models with 59% smaller model sizes than the closest prior work. When deployed on a microcontroller, QUTE demonstrates a 31% reduction in latency on average. In addition, we show that QUTE excels at detecting accuracy-drop events, outperforming all prior works.

The imperative to deploy Deep Neural Network (DNN) models on resource-constrained edge devices, spurred by privacy concerns, has become increasingly apparent. To facilitate the transition from cloud to edge computing, this paper introduces a technique that effectively reduces the memory footprint of DNNs, accommodating the limitations of resource-constrained edge devices while preserving model accuracy. Our proposed technique, named Post-Training Intra-Layer Multi-Precision Quantization (PTILMPQ), employs a post-training quantization approach, eliminating the need for extensive training data. By estimating the importance of layers and channels within the network, the proposed method enables precise bit allocation throughout the quantization process. Experimental results demonstrate that PTILMPQ offers a promising solution for deploying DNNs on edge devices with restricted memory resources. For instance, in the case of ResNet50, it achieves an accuracy of 74.57\% with a memory footprint of 9.5 MB, representing a 25.49\% reduction compared to previous similar methods, with only a minor 1.08\% decrease in accuracy., Comment: The 25th International Symposium on Quality Electronic Design (ISQED'24)

We consider training decision trees using noisily labeled data, focusing on loss functions that can lead to robust learning algorithms. Our contributions are threefold. First, we offer novel theoretical insights on the robustness of many existing loss functions in the context of decision tree learning. We show that some of the losses belong to a class of what we call conservative losses, and the conservative losses lead to an early stopping behavior during training and noise-tolerant predictions during testing. Second, we introduce a framework for constructing robust loss functions, called distribution losses. These losses apply percentile-based penalties based on an assumed margin distribution, and they naturally allow adapting to different noise rates via a robustness parameter. In particular, we introduce a new loss called the negative exponential loss, which leads to an efficient greedy impurity-reduction learning algorithm. Lastly, our experiments on multiple datasets and noise settings validate our theoretical insight and the effectiveness of our adaptive negative exponential loss., Comment: Accepted at AAAI Conference on Artificial Intelligence 2024

Gas Electron Multiplier (GEM) detectors are crucial for enabling high-resolution X-ray polarisation of astrophysical sources when coupled to custom pixel readout ASIC in Gas Pixel Detectors (GPD), as in the Imaging X-ray Polarimetry Explorer (IXPE), the Polarlight cubesat pathfinder and the PFA telescope onboard the future large enhanced X-ray Timing and Polarimetry (eXTP) Chinese mission. The R&D efforts of the IXPE collaboration have resulted in mature GPD technology. However, limitations in the classical wet-etch or laser-drilled fabrication process of GEMs motivated our exploration of alternative methods. This work focuses on investigating a plasma-based etching approach for fabricating GEM patterns at Fondazione Bruno Kessler (FBK). The objective is to improve the aspect ratio of the GEM holes, to mitigate the charging of the GEM dielectric which generates rate-dependent gain changes. Unlike the traditional wet-etch process, Reactive Ion Etching (RIE) enables more vertical etching profiles and thus better aspect ratios. Moreover, the RIE process promises to overcome non-uniformities in the GEM hole patterns which are believed to cause systemic effects in the azimuthal response of GPDs equipped with either laser-drilled or wet-etch GEMs. We present a GEM geometry with 20 micrometers in diameter and 50 micrometers pitch, accompanied by extensive characterisation (SEM and PFIB) of the structural features and aspect ratios. The collaboration with INFN Pisa and Turin enabled us to compare the electrical properties of these detectors and test their performance in their use as electron multipliers in GPDs. Although this R&D work is in its initial stages, it holds promise for enhancing the sensitivity of the IXPE mission in X-ray polarimetry measurements through GEM pattern with more vertical hole profiles., Comment: IPRD2023 conference

Timely and adequate rehabilitation is critical in facilitating post-stroke recovery. However, the organization and delivery of rehabilitation are resource-demanding, and are only available to approximately 25% of stroke survivors in low-to-middle-income countries. Improving access to stroke rehabilitation services through innovative solutions is therefore urgently required. Tele-rehabilitation, which transits care to home- and community settings, has emerged as a promising solution. However, current approaches using video tutorial, teleconference, or other specialized devices face inherent shortfalls that limit their uptake. In this study, we proposed and validated the use of an open-source, markerless motion capture model with consumer-grade devices to overcome these challenges. Our solution enables reliable measurement of the end range of motion during upper limb exercises with near-perfect waveform similarity and intraclass correlation to that of the gold standard Kinect approach. Our multidisciplinary team developed an automated telerehabilitation framework incorporating the validated markerless technique to facilitate a seamless telerehabilitation process. It enables personalized rehabilitation plans with real-time feedback, and individual progress reports using objective quantitative and qualitative features to improve patient monitoring and management, and home-based rehabilitation service uptake and compliance. This study serves as a proof-of-concept in preparation for the future development of a detailed model of care, and feasibility, usability, and cost-effectiveness studies of an automated telerehabilitation platform and framework in improving the state of post-stroke rehabilitation and functional outcome., Comment: it's not a ready paper to be uploaded

Knowledge of the neutron light-yield response is crucial to the understanding of scintillator-based neutron detectors. In this work, neutrons from 2--6 MeV have been used to study the scintillation light-yield response of the liquid scintillators NE 213A, EJ 305, EJ 331 and EJ 321P using event-by-event waveform digitization. Energy calibration was performed using a GEANT model to locate the edge positions of the Compton distributions produced by gamma-ray sources. The simulated light yield for neutrons from a PuBe source was compared to measured recoil proton distributions, where neutron energy was selected by time-of-flight. This resulted in an energy-dependent Birks parametrization to characterize the non-linear response to the lower energy neutrons. The NE 213A and EJ 305 results agree very well with existing data and are reproduced nicely by the simulation. New results for EJ 331 and EJ 321P, where the simulation also reproduces the data well, are presented., Comment: 31 pages, 15 figures, 4 tables

Automatic Adjustment Mechanisms (AAM) are legal instruments that help social security systems respond to demographic and economic changes. In Brazil, the Social Security Factor (SSF) was introduced in the late 1990s as an AAM to link retirement benefits to life expectancy at the retirement age, with the hope of promoting contributory justice and discouraging early retirement. Recent research has highlighted the limitations of right-censored life tables, such as those used in Brazil. It has recommended using the gamma-Gompertz-Makeham (GGM) model to estimate adult and old-age mortality. This study investigated the impact of right-censoring on the SSF by comparing the official SSF and other social security metrics with a counterfactual scenario computed based on fitted GGM models. The results indicate that from 2004 to 2012, official life tables may have negatively impacted retirees' income, particularly for those who delayed their retirement. Furthermore, the GGM-fitted models' life expectancies had more stable paths over time, which could have helped with long-term planning. This study's findings are significant for policymakers as they highlight the importance of using appropriate mortality metrics in AAMs to ensure accurate retirement benefit payments. They also underscore the need to consider the potential impacts of seemingly innocuous hypotheses on public action outcomes. Overall, this study provides valuable insights for public planners and policymakers looking to enhance the effectiveness and fairness of social security systems.

BlueSTEAl, the Blue (aluminum chamber of) Silicon TElescope Arrays for light nuclei,has been developed to study direct reactions in inverse kinematics, as well as scattering and breakup reactions using radioactive ion beams. It is a detector system consisting of a pair of annular silicon detector arrays and a zero-degree phoswich plastic scintillator. For typical binary reaction studies in inverse kinematics, light ions are detected by the Si array in coincidence with heavy recoils detected by the phoswich placed at the focal-plane of a zero-degree magnetic spectrometer. The Si array can also be used to detect light nuclei such as berylium and carbon with clear isotope separation, while the phoswich can also be placed at zero degrees without a spectrometer and used as a high-efficiency beam counting monitor with particle identification capability at the rate of up to 5*10^4 particles per second. This paper reports on the capabilities of BlueSTEAl as determined by recent experiments performed at the Texas A&M Cyclotron Institute. The device is also anticipated to be used in future experiments at other radioactive ion beam facilities.

The publication of the \emph{Gaia} catalogue and improvements in methods to determine memberships and fundamental parameters of open clusters has led to major advances in recent years. However, important parameters such as the masses of these objects, although being studied mostly in some isolated cases, have not been addressed in large homogeneous samples based on \emph{Gaia} data, taking into account details such as binary fractions. Consequently, relevant aspects such as the existence of mass segregation were not adequately studied. Within this context, in this work, we introduce a new method to determine individual stellar masses, including an estimation for the ones in binary systems. This method allows us to study the mass of open clusters, as well as the mass functions of the binary star populations. We validate the method and its efficiency and characterize uncertainties using a grid of synthetic clusters with predetermined parameters. We highlight the application of the method to the Pleiades cluster, showing that the results obtained agree with the current consensus in the literature as well as recent \emph{Gaia} data. We then applied the procedure to a sample of 773 open clusters with fundamental parameters determined using \emph{Gaia Early Data Release 3 (eDR3)} data, obtaining their masses. Subsequently, we investigated the relation between the masses and other fundamental parameters of the clusters. Among the results, we found no significant evidence that clusters in our sample lose and segregate mass with age., Comment: 26 pages, 17 figures, accepted for publication in Monthly Notices of the Royal Astronomical Society

We consider the repair scheme of Guruswami-Wootters for the Reed-Solomon code and ask: can we correctly repair a failed node in the presence of erroneous nodes? Equivalently, we consider the collection of downloaded traces as a code and investigate its code-distance properties. We propose three lower bounds on its minimum distance and study methods to efficiently correct errors close to these bounds., Comment: Accepted to IEEE International Symposium on Information Theory 2023

We develop a framework to solve a large class of linearly driven non-Hermitian quantum systems. Such a class of models in the Hermitian scenario is commonly known as multi-state Landau-Zener models. The non-hermiticity is due to the anti-Hermitian couplings between the diabatic levels. We find that there exists a new conservation law, unique to this class of models, that describes the simultaneous growth of the unnormalized wavefunctions. These models have practical applications in Bose-Einstein condensates, and they can describe the dynamics of multi-species bosonic systems. The conservation law relates to a pair-production mechanism that explains the dissociation of diatomic molecules into atoms. We provide a general framework for both solvable and semiclassically solvable non-Hermitian Landau-Zener models. Our findings will open new avenues for a number of diverse emergent phenomena in explicitly time-dependent non-Hermitian quantum systems.

We consider the problem of evaluating distinct multivariate polynomials over several massive datasets in a distributed computing system with a single master node and multiple worker nodes. We focus on the general case when each multivariate polynomial is evaluated over its corresponding dataset and propose a generalization of the Lagrange Coded Computing framework (Yu et al. 2019) to perform all computations simultaneously while providing robustness against stragglers who do not respond in time, adversarial workers who respond with wrong computation and information-theoretic security of dataset against colluding workers. Our scheme introduces a small computation overhead which results in a reduction in download cost and also offers comparable resistance to stragglers over existing solutions. On top of it, we also propose two verification schemes to detect the presence of adversaries, which leads to incorrect results, without involving additional nodes., Comment: 13 pages, 1 figure, 2 tables

Weakly supervised detection of anomalies in surveillance videos is a challenging task. Going beyond existing works that have deficient capabilities to localize anomalies in long videos, we propose a novel glance and focus network to effectively integrate spatial-temporal information for accurate anomaly detection. In addition, we empirically found that existing approaches that use feature magnitudes to represent the degree of anomalies typically ignore the effects of scene variations, and hence result in sub-optimal performance due to the inconsistency of feature magnitudes across scenes. To address this issue, we propose the Feature Amplification Mechanism and a Magnitude Contrastive Loss to enhance the discriminativeness of feature magnitudes for detecting anomalies. Experimental results on two large-scale benchmarks UCF-Crime and XD-Violence manifest that our method outperforms state-of-the-art approaches.

The European Spallation Source (ESS) in Lund, Sweden will become the world's most powerful thermal neutron source. The Macromolecular Diffractometer (NMX) at the ESS requires three 51.2 x 51.2~cm$^{2}$ detectors with reasonable detection efficiency, sub-mm spatial resolution, a narrow point spread function (PSF) and good time resolution. This work presents measurements with the improved version of the NMX detector prototype consisting of a Triple-GEM detector with natural Gd converter and a low material budget readout. The detector was successfully tested at the neutron reactor of the Budapest Neutron Centre (BNC) and at the D16 instrument at the Institut Laue-Langevin (ILL) in Grenoble. The measurements with Cadmium and Gadolinium masks in Budapest demonstrate that the point spread function of the detector lacks long tails that could impede the measurement of diffraction spot intensities. On the D16 instrument at ILL, diffraction spots from Triose phosphate isomerase w/ 2-phosphoglycolate (PGA) inhibitor were measured both in the D16 Helium-3 detector and the Gd-GEM. The comparison between the two detectors show a similar point spread function in both detectors, and the expected efficiency ratio compared to the Helium-3 detector. Both measurements together thus give good indications that the Gd-GEM detector fits the requirements for the NMX instrument at ESS.

Accurate conversion of neutron time-of-flight (TOF) to wavelength, and its uncertainty, is of fundamental importance to neutron scattering measurements. Especially in cases where instruments are highly configurable, the determination of the absolute wavelength after any change must always be performed. Inspired by the manner with which neutron spectrometers determine the absolute wavelength, we evaluate for the first time, in the author's knowledge, a commonly used method for converting TOF to neutron wavelength, the distance of a monitor from the source of neutrons and we analytically calculate the uncertainty contributions that limit the precision of the conversion. The method was evaluated at the V20 test beamline at the Helmholtz Zentrum Berlin (HZB), emulating the ESS source with a pulse of 2.86 ms length and 14 Hz repetition rate, by using a mini-chopper operated at 140 Hz, beam monitors (BMs) and data acquisition infrastructure. The mini-chopper created well-defined neutron pulses and the BM was placed at two positions, enabling the average wavelength of each of the pulses created to be determined. The used experimental setup resulted in absolute wavelength determination at the monitor positions with a $\delta \lambda_{mean} / \lambda_{mean}$ of $\sim$1.8% for $\lambda >4$ \r{A}. With a modest increase of the distance between the reference monitor positions a $\delta \lambda_{mean} / \lambda_{mean}$ of below 0.5% can be achieved. Further improvements are possible by using a thinner monitor, smaller chopper disc openings and a higher rotational speed chopper. The method requires only two neutron measurements and doesn't necessitate the use of crystals or complex fitting, and could constitute a suitable addition to imaging, diffraction, reflectometers and small angle neutron scattering instruments, at spallation sources, that do not normally utilise fast choppers.

New invariants for 2-dimensional cell complexes are defined, which can be interpreted as curvature bounds. These invariants are proved to be rational and computable in a companion article. This document is a survey that collects theorems about these invariants, computes examples, and lays out a programme of conjectures., Comment: 66 pages, 2 figures, companion to arXiv:2210.09841. v2 minor edits. v3 incorporates referees' comments. This is the final version accepted for publication

Let $X$ be a finite, 2-dimensional cell complex. We show that the curvature invariants $\rho_\pm(X)$ and $\sigma_\pm(X)$ are the extrema of explicit rational linear-programming problems. As a result they are rational, realised, and can be computed algorithmically., Comment: 37 pages, 3 figures, companion to arXiv:2210.09841. v3 corrects an error in Section 2

The need to learn from positive and unlabeled data, or PU learning, arises in many applications and has attracted increasing interest. While random forests are known to perform well on many tasks with positive and negative data, recent PU algorithms are generally based on deep neural networks, and the potential of tree-based PU learning is under-explored. In this paper, we propose new random forest algorithms for PU-learning. Key to our approach is a new interpretation of decision tree algorithms for positive and negative data as \emph{recursive greedy risk minimization algorithms}. We extend this perspective to the PU setting to develop new decision tree learning algorithms that directly minimizes PU-data based estimators for the expected risk. This allows us to develop an efficient PU random forest algorithm, PU extra trees. Our approach features three desirable properties: it is robust to the choice of the loss function in the sense that various loss functions lead to the same decision trees; it requires little hyperparameter tuning as compared to neural network based PU learning; it supports a feature importance that directly measures a feature's contribution to risk minimization. Our algorithms demonstrate strong performance on several datasets. Our code is available at \url{https://github.com/puetpaper/PUExtraTrees}., Comment: Accepted at NeurIPS 2022

Motivated by applications in distributed storage, distributed computing, and homomorphic secret sharing, we study communication-efficient schemes for computing linear combinations of coded symbols. Specifically, we design low-bandwidth schemes that evaluate the weighted sum of $\ell$ coded symbols in a codeword $\pmb{c}\in\mathbb{F}^n$, when we are given access to $d$ of the remaining components in $\pmb{c}$. Formally, suppose that $\mathbb{F}$ is a field extension of $\mathbb{B}$ of degree $t$. Let $\pmb{c}$ be a codeword in a Reed-Solomon code of dimension $k$ and our task is to compute the weighted sum of $\ell$ coded symbols. In this paper, for some $s

We develop the theoretical framework of nonequilibrium ultrafast photonics in monolayer quantum spin-Hall insulators supporting a multitude of topological states. In these materials, ubiquitous strong light-matter interactions in the femtosecond scale lead to non-adiabatic quantum dynamics, resulting in topology-dependent nonlinear optoelectronic transport phenomena. We investigate the mechanism driving topological Dirac fermions interacting with strong ultrashort light pulses and uncover various experimentally accessible physical quantities that encode fingerprints of the quantum material's topological electronic state from the high harmonic generated spectrum. Our work sets the theoretical cornerstones to realize the full potential of time-resolved harmonic spectroscopy for identifying topological invariants in two-dimensional quantum spin-Hall solid state systems., Comment: 4 figures

Deploying Machine learning (ML) on milliwatt-scale edge devices (tinyML) is gaining popularity due to recent breakthroughs in ML and Internet of Things (IoT). Most tinyML research focuses on model compression techniques that trade accuracy (and model capacity) for compact models to fit into the KB-sized tiny-edge devices. In this paper, we show how such models can be enhanced by the addition of an early exit intermediate classifier. If the intermediate classifier exhibits sufficient confidence in its prediction, the network exits early thereby, resulting in considerable savings in time. Although early exit classifiers have been proposed in previous work, these previous proposals focus on large networks, making their techniques suboptimal/impractical for tinyML applications. Our technique is optimized specifically for tiny-CNN sized models. In addition, we present a method to alleviate the effect of network overthinking by leveraging the representations learned by the early exit. We evaluate T-RecX on three CNNs from the MLPerf tiny benchmark suite for image classification, keyword spotting and visual wake word detection tasks. Our results show that T-RecX 1) improves the accuracy of baseline network, 2) achieves 31.58% average reduction in FLOPS in exchange for one percent accuracy across all evaluated models. Furthermore, we show that our methods consistently outperform popular prior works on the tiny-CNNs we evaluate., Comment: Accepted at 20th ACM International Conference on Computing Frontiers

Fast-neutron/gamma-ray pulse-shape discrimination has been performed for the organic liquid scintillators NE 213A and EJ 305 using a time-of-flight based neutron-tagging technique and waveform digitization on an event-by-event basis. Gamma-ray sources and a Geant4-based simulation were used to calibrate the scintillation-light yield. The difference in pulse shape for the neutron and gamma-ray events was analysed by integrating selected portions of the digitized waveform to produce a figure-of-merit for neutron/gamma-ray separation. This figure-of-merit has been mapped as a function of detector threshold and also of neutron energy determined from time-of-flight. It shows clearly that the well-established pulse-shape discrimination capabilities of NE 213A are superior to those of EJ 305. The extra information provided by the neutron-tagging technique has resulted in a far more detailed assessment of the pulse-shape discrimination capabilities of these organic scintillators., Comment: 22 pages, 12 figures, 2 tables

The Helium-3 shortage and the growing interest in neutron science constitute a driving factor in developing new neutron detection technologies. In this work, we report the development of a double-GEM detector prototype that uses a $^{10}$B$_4$C layer as a neutron converter material. GEANT4 simulations were performed predicting an efficiency of 3.14(10) %, agreeing within 2.7 $\sigma$ with the experimental and analytic detection efficiencies obtained by the detector when tested in a 41.8 meV thermal neutron beam. The detector is position sensitive, equipped with a 256+256 strip readout connected to resistive chains, and achieves a spatial resolution better than 3 mm. The gain stability over time was also measured with a fluctuation of about 0.2 %h$^{-1}$ of the signal amplitude. A simple data acquisition with only 5 electronic channels is sufficient to operate this detector., Comment: Prepared for submission to JINST