Searcharxiv⌕ Search

arXiv subjects

Hien Nguyen

Publications and source records attributed to Hien Nguyen.

At least 37 records · Page 2Linked to original sources
arXiv2023

High-sensitivity transition-edge-sensed bolometers: improved speed and characterization with AC and DC bias

We report on efforts to improve the speed of low-G far-infrared transition-edged-sensed bolometers. We use a fabrication process that does not require any dry etch steps to reduce heat capacity on the suspended device and measure a reduction in the detector time constant. However, we also measure an increase in the temperature-normalized thermal conductance (G), and a corresponding increase in the noise-equivalent power (NEP). We employ a new near-IR photon-noise technique using a near-IR laser to calibrate the frequency-domain multiplexed AC system and compare the results to a well-understood DC circuit. We measure an NEP white noise level of 0.8 aW/rtHz with a 1/f knee below 0.1 Hz and a time constant of 3.2 ms.

physics.app-ph↗
arXiv2023

Automated wildlife image classification: An active learning tool for ecological applications

Wildlife camera trap images are being used extensively to investigate animal abundance, habitat associations, and behavior, which is complicated by the fact that experts must first classify the images manually. Artificial intelligence systems can take over this task but usually need a large number of already-labeled training images to achieve sufficient performance. This requirement necessitates human expert labor and poses a particular challenge for projects with few cameras or short durations. We propose a label-efficient learning strategy that enables researchers with small or medium-sized image databases to leverage the potential of modern machine learning, thus freeing crucial resources for subsequent analyses. Our methodological proposal is two-fold: (1) We improve current strategies of combining object detection and image classification by tuning the hyperparameters of both models. (2) We provide an active learning (AL) system that allows training deep learning models very efficiently in terms of required human-labeled training images. We supply a software package that enables researchers to use these methods directly and thereby ensure the broad applicability of the proposed framework in ecological practice. We show that our tuning strategy improves predictive performance. We demonstrate how the AL pipeline reduces the amount of pre-labeled data needed to achieve a specific predictive performance and that it is especially valuable for improving out-of-sample predictive performance. We conclude that the combination of tuning and AL increases predictive performance substantially. Furthermore, we argue that our work can broadly impact the community through the ready-to-use software package provided. Finally, the publication of our models tailored to European wildlife data enriches existing model bases mostly trained on data from Africa and North America.

cs.CV↗
arXiv2022

SPHERExLabTools (SLT): A Python Data Acquisition System for SPHEREx Characterization and Calibration

Selected as the next NASA Medium Class Explorer mission, SPHEREx, the Spectro-Photometer for the History of the Universe, Epoch of Reionization, and Ices Explorer is planned for launch in early 2025. SPHEREx calibration data products include detector spectral response, non-linearity, persistence, and telescope focus error measurements. To produce these calibration products, we have developed a dedicated data acquisition and instrument control system, SPHERExLabTools (SLT). SLT implements driver-level software for control of all testbed instrumentation, graphical interfaces for control of instruments and automated measurements, real-time data visualization, processing, and data archival tools for a variety of output file formats. This work outlines the architecture of the SLT software as a framework for general purpose laboratory data acquisition and instrument control. Initial SPHEREx calibration products acquired while using SLT are also presented.

astro-ph.IM↗
arXiv2022

CapsNet for Medical Image Segmentation

Convolutional Neural Networks (CNNs) have been successful in solving tasks in computer vision including medical image segmentation due to their ability to automatically extract features from unstructured data. However, CNNs are sensitive to rotation and affine transformation and their success relies on huge-scale labeled datasets capturing various input variations. This network paradigm has posed challenges at scale because acquiring annotated data for medical segmentation is expensive, and strict privacy regulations. Furthermore, visual representation learning with CNNs has its own flaws, e.g., it is arguable that the pooling layer in traditional CNNs tends to discard positional information and CNNs tend to fail on input images that differ in orientations and sizes. Capsule network (CapsNet) is a recent new architecture that has achieved better robustness in representation learning by replacing pooling layers with dynamic routing and convolutional strides, which has shown potential results on popular tasks such as classification, recognition, segmentation, and natural language processing. Different from CNNs, which result in scalar outputs, CapsNet returns vector outputs, which aim to preserve the part-whole relationships. In this work, we first introduce the limitations of CNNs and fundamentals of CapsNet. We then provide recent developments of CapsNet for the task of medical image segmentation. We finally discuss various effective network architectures to implement a CapsNet for both 2D images and 3D volumetric medical image segmentation.

eess.IV↗
arXiv2021

Strong Negative Electrothermal Feedback in Thermal Kinetic Inductance Detectors

We demonstrate strong negative electrothermal feedback accelerating and linearizing the response of a thermal kinetic inductance detector (TKID). TKIDs are a proposed highly multiplexable replacement to transition-edge sensors and measure power through the temperature-dependent resonant frequency of a superconducting microresonator bolometer. At high readout probe power and probe frequency detuned from the TKID resonant frequency, we observe electrothermal feedback loop gain up to $\mathcal L$ $\approx$ 16 through measuring the reduction of settling time. We also show that the detector response has no detectable non-linearity over a 38% range of incident power and that the noise-equivalent power is below the design photon noise.

physics.ins-det↗
arXiv2021

Explaining the data or explaining a model? Shapley values that uncover non-linear dependencies

Shapley values have become increasingly popular in the machine learning literature thanks to their attractive axiomatisation, flexibility, and uniqueness in satisfying certain notions of `fairness'. The flexibility arises from the myriad potential forms of the Shapley value \textit{game formulation}. Amongst the consequences of this flexibility is that there are now many types of Shapley values being discussed, with such variety being a source of potential misunderstanding. To the best of our knowledge, all existing game formulations in the machine learning and statistics literature fall into a category which we name the model-dependent category of game formulations. In this work, we consider an alternative and novel formulation which leads to the first instance of what we call model-independent Shapley values. These Shapley values use a (non-parametric) measure of non-linear dependence as the characteristic function. The strength of these Shapley values is in their ability to uncover and attribute non-linear dependencies amongst features. We introduce and demonstrate the use of the energy distance correlations, affine-invariant distance correlation, and Hilbert-Shmidt independence criterion as Shapley value characteristic functions. In particular, we demonstrate their potential value for exploratory data analysis and model diagnostics. We conclude with an interesting expository application to a classical medical survey data set.

stat.ML↗
arXiv2021

Shapley values for feature selection: The good, the bad, and the axioms

The Shapley value has become popular in the Explainable AI (XAI) literature, thanks, to a large extent, to a solid theoretical foundation, including four "favourable and fair" axioms for attribution in transferable utility games. The Shapley value is provably the only solution concept satisfying these axioms. In this paper, we introduce the Shapley value and draw attention to its recent uses as a feature selection tool. We call into question this use of the Shapley value, using simple, abstract "toy" counterexamples to illustrate that the axioms may work against the goals of feature selection. From this, we develop a number of insights that are then investigated in concrete simulation settings, with a variety of Shapley value formulations, including SHapley Additive exPlanations (SHAP) and Shapley Additive Global importancE (SAGE).

cs.LG↗
arXiv2020

Sub-Weibull distributions: generalizing sub-Gaussian and sub-Exponential properties to heavier-tailed distributions

We propose the notion of sub-Weibull distributions, which are characterised by tails lighter than (or equally light as) the right tail of a Weibull distribution. This novel class generalises the sub-Gaussian and sub-Exponential families to potentially heavier-tailed distributions. Sub-Weibull distributions are parameterized by a positive tail index $θ$ and reduce to sub-Gaussian distributions for $θ=1/2$ and to sub-Exponential distributions for $θ=1$. A characterisation of the sub-Weibull property based on moments and on the moment generating function is provided and properties of the class are studied. An estimation procedure for the tail parameter is proposed and is applied to an example stemming from Bayesian deep learning.

math.ST↗
arXiv2020

Evaluation of Inference Attack Models for Deep Learning on Medical Data

Deep learning has attracted broad interest in healthcare and medical communities. However, there has been little research into the privacy issues created by deep networks trained for medical applications. Recently developed inference attack algorithms indicate that images and text records can be reconstructed by malicious parties that have the ability to query deep networks. This gives rise to the concern that medical images and electronic health records containing sensitive patient information are vulnerable to these attacks. This paper aims to attract interest from researchers in the medical deep learning community to this important problem. We evaluate two prominent inference attack models, namely, attribute inference attack and model inversion attack. We show that they can reconstruct real-world medical images and clinical reports with high fidelity. We then investigate how to protect patients' privacy using defense mechanisms, such as label perturbation and model perturbation. We provide a comparison of attack results between the original and the medical deep learning models with defenses. The experimental evaluations show that our proposed defense approaches can effectively reduce the potential privacy leakage of medical deep learning from the inference attacks.

cs.LG↗
arXiv2020

SARGDV: Efficient identification of groundwater-dependent vegetation using synthetic aperture radar

Groundwater depletion impacts the sustainability of numerous groundwater-dependent vegetation (GDV) globally, placing significant stress on their capacity to provide environmental and ecological support for flora, fauna, and anthropic benefits. Industries such as mining, agriculture, and plantations are heavily reliant on groundwater, the over-exploitation of which risks impacting groundwater regimes, quality, and accessibility for nearby GDVs. Cost effective methods of GDV identification will enable strategic protection of these critical ecological systems, through improved and sustainable groundwater management by communities and industry. Recent application of synthetic aperture radar (SAR) earth observation data in Australia has demonstrated the utility of radar for identifying terrestrial groundwater-dependent ecosystems at scale. We propose a robust classification method to advance identification of GDVs at scale using processed SAR data products adapted from a recent previous method. The method includes the development of SARGDV, a binary classification model, which uses the extreme gradient boosting (XGBoost) algorithm in conjunction with three data cubes composed of Sentinel-1 SAR interferometric wide images. The images were collected as a one-year time series over Mount Gambier, a region in South Australia, known to support GDVs. The SARGDV model demonstrated high performance for classifying GDVs with 77% precision, 76% true positive rate and 96% accuracy. This method may be used to support the protection of GDV communities globally by providing a long term, cost-effective solution to identify GDVs over variable regions and climates, via the use of freely available, high-resolution, globally available Sentinel-1 SAR data sets.

stat.AP↗
arXiv2020

Shapley value confidence intervals for attributing variance explained

The coefficient of determination, the $R^2$, is often used to measure the variance explained by an affine combination of multiple explanatory covariates. An attribution of this explanatory contribution to each of the individual covariates is often sought in order to draw inference regarding the importance of each covariate with respect to the response phenomenon. A recent method for ascertaining such an attribution is via the game theoretic Shapley value decomposition of the coefficient of determination. Such a decomposition has the desirable efficiency, monotonicity, and equal treatment properties. Under a weak assumption that the joint distribution is pseudo-elliptical, we obtain the asymptotic normality of the Shapley values. We then utilize this result in order to construct confidence intervals and hypothesis tests regarding such quantities. Monte Carlo studies regarding our results are provided. We found that our asymptotic confidence intervals are computationally superior to competing bootstrap methods and are able to improve upon the performance of such intervals. In an expository application to Australian real estate price modelling, we employ Shapley value confidence intervals to identify significant differences between the explanatory contributions of covariates, between models, which otherwise share approximately the same $R^2$ value. These different models are based on real estate data from the same periods in 2019 and 2020, the latter covering the early stages of the arrival of the novel coronavirus, COVID-19.

stat.ME↗
arXiv2020

Thermal Kinetic Inductance Detectors for millimeter-wave detection

Thermal Kinetic Inductance Detectors (TKIDs) combine the excellent noise performance of traditional bolometers with a radio frequency multiplexing architecture that enables the large detector counts needed for the next generation of millimeter-wave instruments. In this paper, we first discuss the expected noise sources in TKIDs and derive the limits where the phonon noise contribution dominates over the other detector noise terms: generation-recombination, amplifier, and two-level system (TLS) noise. Second, we characterize aluminum TKIDs in a dark environment. We present measurements of TKID resonators with quality factors of about $10^5$ at 80 mK. We also discuss the bolometer thermal conductance, heat capacity, and time constants. These were measured by the use of a resistor on the thermal island to excite the bolometers. These dark aluminum TKIDs demonstrate a noise equivalent power NEP = $2 \times 10^{-17} \mathrm{W}/\mathrm{\sqrt{Hz}} $, with a $1/f$ knee at 0.1 Hz, which provides background noise limited performance for ground-based telescopes observing at 150 GHz.

physics.ins-det↗
arXiv2020

DVNet: A Memory-Efficient Three-Dimensional CNN for Large-Scale Neurovascular Reconstruction

Maps of brain microarchitecture are important for understanding neurological function and behavior, including alterations caused by chronic conditions such as neurodegenerative disease. Techniques such as knife-edge scanning microscopy (KESM) provide the potential for whole organ imaging at sub-cellular resolution. However, multi-terabyte data sizes make manual annotation impractical and automatic segmentation challenging. Densely packed cells combined with interconnected microvascular networks are a challenge for current segmentation algorithms. The massive size of high-throughput microscopy data necessitates fast and largely unsupervised algorithms. In this paper, we investigate a fully-convolutional, deep, and densely-connected encoder-decoder for pixel-wise semantic segmentation. The excessive memory complexity often encountered with deep and dense networks is mitigated using skip connections, resulting in fewer parameters and enabling a significant performance increase over prior architectures. The proposed network provides superior performance for semantic segmentation problems applied to open-source benchmarks. We finally demonstrate our network for cellular and microvascular segmentation, enabling quantitative metrics for organ-scale neurovascular analysis.

eess.IV↗
arXiv2019

Fast and Accurate Capitalization and Punctuation for Automatic Speech Recognition Using Transformer and Chunk Merging

In recent years, studies on automatic speech recognition (ASR) have shown outstanding results that reach human parity on short speech segments. However, there are still difficulties in standardizing the output of ASR such as capitalization and punctuation restoration for long-speech transcription. The problems obstruct readers to understand the ASR output semantically and also cause difficulties for natural language processing models such as NER, POS and semantic parsing. In this paper, we propose a method to restore the punctuation and capitalization for long-speech ASR transcription. The method is based on Transformer models and chunk merging that allows us to (1), build a single model that performs punctuation and capitalization in one go, and (2), perform decoding in parallel while improving the prediction accuracy. Experiments on British National Corpus showed that the proposed approach outperforms existing methods in both accuracy and decoding speed.

cs.CL↗
arXiv2018

Astronomical random numbers for quantum foundations experiments

Photons from distant astronomical sources can be used as a classical source of randomness to improve fundamental tests of quantum nonlocality, wave-particle duality, and local realism through Bell's inequality and delayed-choice quantum eraser tests inspired by Wheeler's cosmic-scale Mach-Zehnder interferometer gedankenexperiment. Such sources of random numbers may also be useful for information-theoretic applications such as key distribution for quantum cryptography. Building on the design of an "astronomical random-number generator" developed for the recent "cosmic Bell" experiment [Handsteiner et al., Phys. Rev. Lett. 118, 060401 (2017)], in this paper we report on the design and characterization of a device that, with 20-nanosecond latency, outputs a bit based on whether the wavelength of an incoming photon is greater than or less than 700 nm. Using the one-meter telescope at the Jet Propulsion Laboratory (JPL) Table Mountain Observatory, we generated random bits from astronomical photons in both color channels from 50 stars of varying color and magnitude, and from 12 quasars with redshifts up to $z = 3.9$. With stars, we achieved bit rates of $\sim 1 \times 10^6$ Hz / m$^2$, limited by saturation for our single-photon detectors, and with quasars of magnitudes between 12.9 and 16, we achieved rates between $\sim 10^2$ and $2 \times 10^3$ Hz /m$^2$. For bright quasars, the resulting bitstreams exhibit sufficiently low amounts of statistical predictability as quantified by the mutual information. In addition, a sufficiently high fraction of bits generated are of true astronomical origin in order to address both the locality and freedom-of-choice loopholes when used to set the measurement settings in a test of the Bell-CHSH inequality.

quant-ph↗
arXiv2018

Testing the Weak Equivalence Principle using Optical and Near-Infrared Crab Pulses

The Weak Equivalence Principle states that the geodesics of a test particle in a gravitational field are independent of the particle's constitution. To constrain violations of the Weak Equivalence Principle, we use the one-meter telescope at Table Mountain Observatory near Los Angeles to monitor the relative arrival times of pulses from the Crab Pulsar in the optical ($λ\approx 585$ nm) and near-infrared ($λ\approx 814$ nm) using an instrument which detects single photons with nanosecond-timing resolution in those two bands. The infrared pulse arrives slightly before the visible pulse. Our three analysis methods give delays with statistical errors of $Δt_{obs} = 7.41 \pm 0.58$, $0.4 \pm 3.6$, and $7.35 \pm 4.48$ microseconds (at most 1/4000 of the pulsar period). We attribute this discrepancy to systematic error from the fact that the visible and infrared pulses have slightly different shapes. Whether this delay emerges from the pulsar, is caused by passing through wavelength-dependent media, or is caused by a violation of the equivalence principle, unless there is a fine-tuned cancellation among these, we set the first upper limit on the differential post-Newtonian parameter at these wavelengths of $Δγ< 1.07 \times 10^{-10}~(3σ)$. This result falls in an unexplored region of parameter space and complements existing limits on equivalence-principle violation from fast radio bursts, gamma ray bursts, as well as previous limits from the Crab.

astro-ph.HE↗
arXiv2016

Science Impacts of the SPHEREx All-Sky Optical to Near-Infrared Spectral Survey: Report of a Community Workshop Examining Extragalactic, Galactic, Stellar and Planetary Science

SPHEREx is a proposed SMEX mission selected for Phase A. SPHEREx will carry out the first all-sky spectral survey and provide for every 6.2" pixel a spectra between 0.75 and 4.18 $μ$m [with R$\sim$41.4] and 4.18 and 5.00 $μ$m [with R$\sim$135]. The SPHEREx team has proposed three specific science investigations to be carried out with this unique data set: cosmic inflation, interstellar and circumstellar ices, and the extra-galactic background light. It is readily apparent, however, that many other questions in astrophysics and planetary sciences could be addressed with the SPHEREx data. The SPHEREx team convened a community workshop in February 2016, with the intent of enlisting the aid of a larger group of scientists in defining these questions. This paper summarizes the rich and varied menu of investigations that was laid out. It includes studies of the composition of main belt and Trojan/Greek asteroids; mapping the zodiacal light with unprecedented spatial and spectral resolution; identifying and studying very low-metallicity stars; improving stellar parameters in order to better characterize transiting exoplanets; studying aliphatic and aromatic carbon-bearing molecules in the interstellar medium; mapping star formation rates in nearby galaxies; determining the redshift of clusters of galaxies; identifying high redshift quasars over the full sky; and providing a NIR spectrum for most eROSITA X-ray sources. All of these investigations, and others not listed here, can be carried out with the nominal all-sky spectra to be produced by SPHEREx. In addition, the workshop defined enhanced data products and user tools which would facilitate some of these scientific studies. Finally, the workshop noted the high degrees of synergy between SPHEREx and a number of other current or forthcoming programs, including JWST, WFIRST, Euclid, GAIA, K2/Kepler, TESS, eROSITA and LSST.

astro-ph.CO↗