arXiv subjects

Hoang Nguyen

Publications and source records attributed to Hoang Nguyen.

31 records · Page 2Linked to original sources

arXiv2023

LVM-Med: Learning Large-Scale Self-Supervised Vision Models for Medical Imaging via Second-order Graph Matching

Obtaining large pre-trained models that can be fine-tuned to new tasks with limited annotated samples has remained an open challenge for medical imaging data. While pre-trained deep networks on ImageNet and vision-language foundation models trained on web-scale data are prevailing approaches, their effectiveness on medical tasks is limited due to the significant domain shift between natural and medical images. To bridge this gap, we introduce LVM-Med, the first family of deep networks trained on large-scale medical datasets. We have collected approximately 1.3 million medical images from 55 publicly available datasets, covering a large number of organs and modalities such as CT, MRI, X-ray, and Ultrasound. We benchmark several state-of-the-art self-supervised algorithms on this dataset and propose a novel self-supervised contrastive learning algorithm using a graph-matching formulation. The proposed approach makes three contributions: (i) it integrates prior pair-wise image similarity metrics based on local and global information; (ii) it captures the structural constraints of feature embeddings through a loss function constructed via a combinatorial graph-matching objective; and (iii) it can be trained efficiently end-to-end using modern gradient-estimation techniques for black-box solvers. We thoroughly evaluate the proposed LVM-Med on 15 downstream medical tasks ranging from segmentation and classification to object detection, and both for the in and out-of-distribution settings. LVM-Med empirically outperforms a number of state-of-the-art supervised, self-supervised, and foundation models. For challenging tasks such as Brain Tumor Classification or Diabetic Retinopathy Grading, LVM-Med improves previous vision-language models trained on 1 billion masks by 6-7% while using only a ResNet-50.

cs.CV↗

arXiv2022

Joint Self-Supervised Image-Volume Representation Learning with Intra-Inter Contrastive Clustering

Collecting large-scale medical datasets with fully annotated samples for training of deep networks is prohibitively expensive, especially for 3D volume data. Recent breakthroughs in self-supervised learning (SSL) offer the ability to overcome the lack of labeled training samples by learning feature representations from unlabeled data. However, most current SSL techniques in the medical field have been designed for either 2D images or 3D volumes. In practice, this restricts the capability to fully leverage unlabeled data from numerous sources, which may include both 2D and 3D data. Additionally, the use of these pre-trained networks is constrained to downstream tasks with compatible data dimensions. In this paper, we propose a novel framework for unsupervised joint learning on 2D and 3D data modalities. Given a set of 2D images or 2D slices extracted from 3D volumes, we construct an SSL task based on a 2D contrastive clustering problem for distinct classes. The 3D volumes are exploited by computing vectored embedding at each slice and then assembling a holistic feature through deformable self-attention mechanisms in Transformer, allowing incorporating long-range dependencies between slices inside 3D volumes. These holistic features are further utilized to define a novel 3D clustering agreement-based SSL task and masking embedding prediction inspired by pre-trained language models. Experiments on downstream tasks, such as 3D brain segmentation, lung nodule detection, 3D heart structures segmentation, and abnormal chest X-ray detection, demonstrate the effectiveness of our joint 2D and 3D SSL approach. We improve plain 2D Deep-ClusterV2 and SwAV by a significant margin and also surpass various modern 2D and 3D SSL approaches.

cs.CV↗

arXiv2022

Monitoring the Dynamic Networks of Stock Returns

In this paper, we study the connection between the companies in the Swedish capital market. We consider 28 companies included in the determination of the market index OMX30. The network structure of the market is constructed using different methods to determine the distance between the companies. We use hierarchical clustering methods to find the relation among the companies in each window. Next, we obtain one-dimensional time series of the distances between the clustering trees that reflect the changes in the relationship between the companies in the market over time. The method of statistical process control, namely the Shewhart control chart, is applied to those time series to detect abnormal changes in the financial market.

q-fin.ST↗

arXiv2022

Fall detection using multimodal data

In recent years, the occurrence of falls has increased and has had detrimental effects on older adults. Therefore, various machine learning approaches and datasets have been introduced to construct an efficient fall detection algorithm for the social community. This paper studies the fall detection problem based on a large public dataset, namely the UP-Fall Detection Dataset. This dataset was collected from a dozen of volunteers using different sensors and two cameras. We propose several techniques to obtain valuable features from these sensors and cameras and then construct suitable models for the main problem. The experimental results show that our proposed methods can bypass the state-of-the-art methods on this dataset in terms of accuracy, precision, recall, and F1 score.

cs.CV↗

arXiv2022

Exploring Mechanisms of Hydration and Carbonation of MgO and Mg(OH)2 in Reactive Magnesium Oxide-based Cements

Reactive magnesium oxide (MgO)-based cement (RMC) can play a key role in carbon capture processes. However, knowledge on the driving forces that control the degree of carbonation and hydration and rate of reactions in this system remains limited. In this work, density functional theory-based simulations are used to investigate the physical nature of the reactions taking place during the fabrication of RMCs under ambient conditions. Parametric indicators such as adsorption energies, charge transfer, electron localization function, adsorption/dissociation energy barriers and the mechanisms of interaction of H2O and CO2 molecules with MgO and brucite (Mg(OH)2) clusters are considered. The following hydration and carbonation interactions relevant to RMCs are evaluated i) carbonation of MgO, ii) hydration of MgO, carbonation of hydrated MgO, iii) carbonation of Mg(OH)2, iv) hydration of Mg(OH)2 and v) hydration of carbonated Mg(OH)2. A comparison of the energy barriers and reaction pathways of these mechanisms shows that the carbonation of MgO is hindered by presence of H2O molecules, while the carbonation of Mg(OH)2 is hindered by the formation of initial carbonate and hydrate layers as well as presence of excessed H2O molecules. To compare these finding to bulk mineral surfaces, the interactions of the CO2 and H2O molecules with the MgO(001) and Mg(OH)2 (001) surfaces are studied. Therefore, this work presents deep insights into the physical nature of the reactions and the mechanisms involved in hydrated magnesium carbonates production that can be beneficial for its development.

cond-mat.mtrl-sci↗

arXiv2021

Vector autoregression models with skewness and heavy tails

With uncertain changes of the economic environment, macroeconomic downturns during recessions and crises can hardly be explained by a Gaussian structural shock. There is evidence that the distribution of macroeconomic variables is skewed and heavy tailed. In this paper, we contribute to the literature by extending a vector autoregression (VAR) model to account for a more realistic assumption of the multivariate distribution of the macroeconomic variables. We propose a general class of generalized hyperbolic skew Student's t distribution with stochastic volatility for the error term in the VAR model that allows us to take into account skewness and heavy tails. Tools for Bayesian inference and model selection using a Gibbs sampler are provided. In an empirical study, we present evidence of skewness and heavy tails for monthly macroeconomic variables. The analysis also gives a clear message that skewness should be taken into account for better predictions during recessions and crises.

econ.EM↗

arXiv2020

Solid shell prism elements based on hierarchical, heterogeneous, and anisotropic shape functions

The formulation of a new prism finite element is presented for the nonlinear analysis of solid shells subject to large strains and large displacements. The element is based on hierarchical, heterogeneous, and anisotropic shape functions. As with other solid shell formulations, only displacement degrees of freedom are required to describe the shell kinematics and general three-dimensional material laws can be adopted. However, the novelty of this formulation is the ability to capture complex shell behaviour and avoid locking phenomena, without the need to use reduced integration or adopt additional natural strain or enhanced strain fields. Thus, this element is ideally suited for geometrically and physically nonlinear problems. This is achieved by constructing independent approximation shape functions on both the prism element's triangular faces and through the thickness, where the latter is associated with a local coordinate system that convects with deformation of the shell. The element is extremely efficient, with the hierarchical property lending itself to an efficient and highly scalable multigrid solver, and the heterogeneity property enables local p-adaptivity. The paper demonstrates performance of the element for a number of linear and geometrically nonlinear problems, benchmarked against well established problems in the literature. The formulation has been implemented in the MoFEM.

math.NA↗

arXiv2020

Dynamic Semantic Matching and Aggregation Network for Few-shot Intent Detection

Few-shot Intent Detection is challenging due to the scarcity of available annotated utterances. Although recent works demonstrate that multi-level matching plays an important role in transferring learned knowledge from seen training classes to novel testing classes, they rely on a static similarity measure and overly fine-grained matching components. These limitations inhibit generalizing capability towards Generalized Few-shot Learning settings where both seen and novel classes are co-existent. In this paper, we propose a novel Semantic Matching and Aggregation Network where semantic components are distilled from utterances via multi-head self-attention with additional dynamic regularization constraints. These semantic components capture high-level information, resulting in more effective matching between instances. Our multi-perspective matching method provides a comprehensive matching measure to enhance representations of both labeled and unlabeled instances. We also propose a more challenging evaluation setting that considers classification on the joint all-class label space. Extensive experimental results demonstrate the effectiveness of our method. Our code and data are publicly available.

cs.CL↗

arXiv2020

CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection

In this paper, we formulate a more realistic and difficult problem setup for the intent detection task in natural language understanding, namely Generalized Few-Shot Intent Detection (GFSID). GFSID aims to discriminate a joint label space consisting of both existing intents which have enough labeled data and novel intents which only have a few examples for each class. To approach this problem, we propose a novel model, Conditional Text Generation with BERT (CG-BERT). CG-BERT effectively leverages a large pre-trained language model to generate text conditioned on the intent label. By modeling the utterance distribution with variational inference, CG-BERT can generate diverse utterances for the novel intents even with only a few utterances available. Experimental results show that CG-BERT achieves state-of-the-art performance on the GFSID task with 1-shot and 5-shot settings on two real-world datasets.

cs.CL↗

arXiv2018

Branching of Hydraulic Cracks in Gas or Oil Shale with Closed Natural Fractures: How to Master Permeability

While the hydraulic fracturing technology, aka fracking (or fraccing, frac), has become highly developed and astonishingly successful, a consistent formulation of the associated fracture mechanics that would not conflict with some observations is still unavailable. It is attempted here. Classical fracture mechanics, as well as the current commercial softwares, predict vertical cracks to propagate without branching from the perforations of the horizontal well casing, which are typically spaced at 10 m or more. However, to explain the gas production rate at the wellhead, the crack spacing would have to be only about 0.1 m, which would increase the overall gas permeability of shale mass about 10,000$\times$. This permeability increase has generally been attributed to a preexisting system of orthogonal natural cracks, whose spacing is about 0.1 m. But their average age is about 100 million years, and a recent analysis indicated that these cracks must have been completely closed by secondary creep of shale in less than a million years. Here it is considered that the tectonic events that produced the natural cracks in shale must have also created weak layers with nano- or micro-cracking damage. It is numerically demonstrated that a greatly enhanced permeability along the weak layers, with a greatly increased transverse Biot coefficient, must cause the fracking to engender lateral branching and the opening of hydraulic cracks along the weak layers, even if these cracks are initially almost closed. A finite element crack band model, based on recently developed anisotropic spherocylindrical microplane constitutive law, demonstrates these findings.

physics.geo-ph↗

arXiv2018

Question-Answer Selection in User to User Marketplace Conversations

Sellers in user to user marketplaces can be inundated with questions from potential buyers. Answers are often already available in the product description. We collected a dataset of around 590K such questions and answers from conversations in an online marketplace. We propose a question answering system that selects a sentence from the product description using a neural-network ranking model. We explore multiple encoding strategies, with recurrent neural networks and feed-forward attention layers yielding good results. This paper presents a demo to interactively pose buyer questions and visualize the ranking scores of product description sentences from live online listings.

cs.CL↗

arXiv2016

Business Process Deviance Mining: Review and Evaluation

Business process deviance refers to the phenomenon whereby a subset of the executions of a business process deviate, in a negative or positive way, with respect to its expected or desirable outcomes. Deviant executions of a business process include those that violate compliance rules, or executions that undershoot or exceed performance targets. Deviance mining is concerned with uncovering the reasons for deviant executions by analyzing business process event logs. This article provides a systematic review and comparative evaluation of deviance mining approaches based on a family of data mining techniques known as sequence classification. Using real-life logs from multiple domains, we evaluate a range of feature types and classification methods in terms of their ability to accurately discriminate between normal and deviant executions of a process. We also analyze the interestingness of the rule sets extracted using different methods. We observe that feature sets extracted using pattern mining techniques only slightly outperform simpler feature sets based on counts of individual activity occurrences in a trace.

cs.AI↗

arXiv2008

Dislocation interaction with C in alpha-Fe: a comparison between atomic simulations and elasticity theory

The interaction of C atoms with a screw and an edge dislocation is modelled at an atomic scale using an empirical Fe-C interatomic potential based on the Embedded Atom Method (EAM) and molecular statics simulations. Results of atomic simulations are compared with predictions of elasticity theory. It is shown that a quantitative agreement can be obtained between both modelling techniques as long as anisotropic elastic calculations are performed and both the dilatation and the tetragonal distortion induced by the C interstitial are considered. Using isotropic elasticity allows to predict the main trends of the interaction and considering only the interstitial dilatation will lead to a wrong interaction.

cond-mat.mtrl-sci↗