arXiv subjects

Junwei Ma

Publications and source records attributed to Junwei Ma.

At least 37 records · Page 2Linked to original sources

arXiv2024

Bright dipolar excitons in twisted black phosphorus homostructures

Bright dipolar excitons, which contain electrical dipoles and have high oscillator strength, are an ideal platform for studying correlated quantum phenomena. They usually rely on carrier tunneling between two quantum wells or two layers to hybridize with nondipolar excitons to gain oscillator strength. In this work, we uncovered a new type of bright infrared dipolar exciton by stacking 90°-twisted black phosphorus (BP) structures. These excitons, inherent to the reconstructed band structure, exhibit high oscillator strength. Most importantly, they inherit the linear polarization from BP, which allows light polarization to be used to select the dipole direction. Moreover, the dipole moment and resonance energy can be widely tuned by the thickness of the BP. Our results demonstrate a useful platform for exploring tunable correlated dipolar excitons.

cond-mat.mes-hall↗

arXiv2024

Establishing Nationwide Power System Vulnerability Index across US Counties Using Interpretable Machine Learning

Power outages have become increasingly frequent, intense, and prolonged in the US due to climate change, aging electrical grids, and rising energy demand. However, largely due to the absence of granular spatiotemporal outage data, we lack data-driven evidence and analytics-based metrics to quantify power system vulnerability. This limitation has hindered the ability to effectively evaluate and address vulnerability to power outages in US communities. Here, we collected ~179 million power outage records at 15-minute intervals across 3022 US contiguous counties (96.15% of the area) from 2014 to 2023. We developed a power system vulnerability assessment framework based on three dimensions (intensity, frequency, and duration) and applied interpretable machine learning models (XGBoost and SHAP) to compute Power System Vulnerability Index (PSVI) at the county level. Our analysis reveals a consistent increase in power system vulnerability over the past decade. We identified 318 counties across 45 states as hotspots for high power system vulnerability, particularly in the West Coast (California and Washington), the East Coast (Florida and the Northeast area), the Great Lakes megalopolis (Chicago-Detroit metropolitan areas), and the Gulf of Mexico (Texas). Heterogeneity analysis indicates that urban counties, counties with interconnected grids, and states with high solar generation exhibit significantly higher vulnerability. Our results highlight the significance of the proposed PSVI for evaluating the vulnerability of communities to power outages. The findings underscore the widespread and pervasive impact of power outages across the country and offer crucial insights to support infrastructure operators, policymakers, and emergency managers in formulating policies and programs aimed at enhancing the resilience of the US power infrastructure.

cs.CY↗

arXiv2024

Recent Decade's Power Outage Data Reveals the Increasing Vulnerability of U.S. Power Infrastructure

Despite significant anecdotal evidence regarding the vulnerability of the U.S. power infrastructure, there is a dearth of longitudinal and nation-level characterization of the spatial and temporal patterns in the frequency and extent of power outages. A data-driven national-level characterization of power outage vulnerability is particularly essential for understanding the urgency and formulating policies to promote the resilience of power infrastructure systems. Recognizing this, we retrieved 179,053,397 county-level power outage records with a 15-minute interval across 3,022 US counties during 2014-2023 to capture power outage characteristics. We focus on three dimensions--power outage intensity, frequency, and duration--and develop multiple metrics to quantify each dimension of power outage vulnerability. The results show that in the past ten years, the vulnerability of U.S. power system has consistently been increasing. Counties experienced an average of 999.4 outages over the decade, affecting an average of more than 540,000 customers per county, with disruptions occurring approximately every week. Coastal areas, particularly in California, Florida and New Jersey, faced more frequent and prolonged outages, while inland regions showed higher outage rates. A concerning increase in outage frequency and intensity was noted, especially after 2017, with a sharp rise in prolonged outages since 2019. The research also found positive association between social vulnerability and outage metrics, with the association becoming stronger over the years under study. Areas with higher social vulnerability experienced more severe and frequent outages, exacerbating challenges in these regions. These findings reveal the much-needed empirical evidence for stakeholders to inform policy formulation and program development for enhancing the resilience of the U.S. power infrastructure.

physics.soc-ph↗

arXiv2024

Retrieval & Fine-Tuning for In-Context Tabular Models

Tabular data is a pervasive modality spanning a wide range of domains, and the inherent diversity poses a considerable challenge for deep learning. Recent advancements using transformer-based in-context learning have shown promise on smaller and less complex datasets, but have struggled to scale to larger and more complex ones. To address this limitation, we propose a combination of retrieval and fine-tuning: we can adapt the transformer to a local subset of the data by collecting nearest neighbours, and then perform task-specific fine-tuning with this retrieved set of neighbours in context. Using TabPFN as the base model -- currently the best tabular in-context learner -- and applying our retrieval and fine-tuning scheme on top results in what we call a locally-calibrated PFN, or LoCalPFN. We conduct extensive evaluation on 95 datasets curated by TabZilla from OpenML, upon which we establish a new state-of-the-art with LoCalPFN -- even with respect to tuned tree-based models. Notably, we show a significant boost in performance compared to the base in-context model, demonstrating the efficacy of our approach and advancing the frontier of deep learning in tabular data.

cs.LG↗

arXiv2024

TabPFGen -- Tabular Data Generation with TabPFN

Advances in deep generative modelling have not translated well to tabular data. We argue that this is caused by a mismatch in structure between popular generative models and discriminative models of tabular data. We thus devise a technique to turn TabPFN -- a highly performant transformer initially designed for in-context discriminative tabular tasks -- into an energy-based generative model, which we dub TabPFGen. This novel framework leverages the pre-trained TabPFN as part of the energy function and does not require any additional training or hyperparameter tuning, thus inheriting TabPFN's in-context learning capability. We can sample from TabPFGen analogously to other energy-based models. We demonstrate strong results on standard generative modelling tasks, including data augmentation, class-balancing, and imputation, unlocking a new frontier of tabular data generation.

cs.LG↗

arXiv2024

Non-locality and Spillover Effects of Residential Flood Damage on Community Recovery: Insights from High-resolution Flood Claim and Mobility Data

Examining the relationship between vulnerability of the built environment and community recovery is crucial for understanding disaster resilience. Yet, this relationship is rather neglected in the existing literature due to previous limitations in the availability of empirical datasets needed for such analysis. In this study, we combine fine-resolution flood damage claims data (composed of both insured and uninsured losses) and human mobility data (composed of millions of movement trajectories) during the 2017 Hurricane Harvey in Harris County, Texas, to specify the extent to which vulnerability of the built environment (i.e., flood property damage) affects community recovery (based on the speed of human mobility recovery) locally and regionally. We examine this relationship using a spatial lag, spatial reach, and spatial decay models to measure the extent of spillover effects of residential damage on community recovery. The findings show that: first, the severity of residential damage significantly affects the speed of community recovery. A greater extent of residential damage suppresses community recovery not only locally but also in the surrounding areas. Second, the spatial spillover effect of residential damage on community recovery speed decays with distance from the highly damaged areas. Third, spatial areas display heterogeneous spatial decay coefficients, which are associated with urban structure features such as the density of points-of-interest facilities and roads. These findings provide a novel data-driven characterization of the spatial diffusion of residential flood damage effects on community recovery and move us closer to a better understanding of complex spatial processes that shape community resilience to hazards. This study also provides valuable insights for emergency managers and public officials seeking to mitigate the non-local effects of residential damage.

stat.AP↗

arXiv2024

Tabular Data Contrastive Learning via Class-Conditioned and Feature-Correlation Based Augmentation

Contrastive learning is a model pre-training technique by first creating similar views of the original data, and then encouraging the data and its corresponding views to be close in the embedding space. Contrastive learning has witnessed success in image and natural language data, thanks to the domain-specific augmentation techniques that are both intuitive and effective. Nonetheless, in tabular domain, the predominant augmentation technique for creating views is through corrupting tabular entries via swapping values, which is not as sound or effective. We propose a simple yet powerful improvement to this augmentation technique: corrupting tabular data conditioned on class identity. Specifically, when corrupting a specific tabular entry from an anchor row, instead of randomly sampling a value in the same feature column from the entire table uniformly, we only sample from rows that are identified to be within the same class as the anchor row. We assume the semi-supervised learning setting, and adopt the pseudo labeling technique for obtaining class identities over all table rows. We also explore the novel idea of selecting features to be corrupted based on feature correlation structures. Extensive experiments show that the proposed approach consistently outperforms the conventional corruption method for tabular data classification tasks. Our code is available at https://github.com/willtop/Tabular-Class-Conditioned-SSL.

cs.LG↗

arXiv2024

Decoding the Pulse of Community during Disasters: Resilience Analysis Based on Fluctuations in Latent Lifestyle Signatures within Human Visitation Networks

Examining the impact of disasters on life activities of populations is critical for understanding community resilience dynamics, yet it remains insufficiently studied in the existing literature. In this study, we leveraged data from more than 1.2 million anonymized human mobility communications across 30 parishes in Louisiana to construct a temporal network that tracks visitation to places from which we characterized human lifestyle signatures before, during, and after Hurricane Ida in 2021. Utilizing the motif model, we distilled complex human lifestyles into identifiable patterns and clustered them into classes: commute, healthcare, dining out, and youth-oriented lifestyle. We defined two metrics to evaluate disruption and recovery fluctuations in lifestyle patterns during the perturbation period compared to the steady period: 1) frequency (daily number of motifs), and 2) proximity (daily average distance of motifs). The results indicate significant dynamics in lifestyle patterns due to the hurricane, with essential facilities (e.g., healthcare) demonstrating a swift recovery. The study underscores the heterogeneity of locations visited and the necessity of integrating both essential and non-essential facilities into disaster response initiatives. Furthermore, our study reveals sustained changes in lifestyle patterns, highlighting the long-term impact of the hurricane on daily life. These insights demonstrate the significance of examining lifestyle signatures and their fluctuations in evaluating disaster resilience patterns for affected communities. The outcomes of this study are poised to aid emergency managers and public officials to more effectively evaluate and monitor disaster impacts and recovery based on changes in lifestyle patterns in the community.

cs.SI↗

arXiv2024

In-Context Data Distillation with TabPFN

Foundation models have revolutionized tasks in computer vision and natural language processing. However, in the realm of tabular data, tree-based models like XGBoost continue to dominate. TabPFN, a transformer model tailored for tabular data, mirrors recent foundation models in its exceptional in-context learning capability, being competitive with XGBoost's performance without the need for task-specific training or hyperparameter tuning. Despite its promise, TabPFN's applicability is hindered by its data size constraint, limiting its use in real-world scenarios. To address this, we present in-context data distillation (ICD), a novel methodology that effectively eliminates these constraints by optimizing TabPFN's context. ICD efficiently enables TabPFN to handle significantly larger datasets with a fixed memory budget, improving TabPFN's quadratic memory complexity but at the cost of a linear number of tuning steps. Notably, TabPFN, enhanced with ICD, demonstrates very strong performance against established tree-based models and modern deep learning methods on 48 large tabular datasets from OpenML.

cs.LG↗

arXiv2023

Layer-dependent exciton polarizability and the brightening of dark excitons in few-layer black phosphorus

The evolution of excitons from 2D to 3D is of great importance in photo-physics, yet the layer-dependent exciton polarizability has not been investigated in 2D semiconductors. Here, we determine the exciton polarizabilities for 3- to 11-layer black phosphorus-a direct bandgap semiconductor regardless of the thickness-through frequency-resolved photocurrent measurements on dual-gate devices and unveil the carrier screening effect in relatively thicker samples. By taking advantage of the broadband photocurrent spectra, we are also able to reveal the exciton response for higher-index subbands under the gate electrical field. Surprisingly, dark excitons are brightened with intensity even stronger than the allowed transitions above certain electrical field. Our study not only sheds light on the exciton evolution with sample thickness, but also paves a way for optoelectronic applications of few-layer BP in modulators, tunable photodetectors, emitters and lasers.

cond-mat.mes-hall↗

arXiv2023

Twist-angle and thickness-ratio tuning of plasmon polaritons in twisted bilayer van der Waals films

Stacking bilayer structures is an efficient way to tune the topology of polaritons in in-plane anisotropic films, e.g., by leveraging the twist angle (TA). However, the effect of another geometric parameter, film thickness ratio (TR), on manipulating the plasmon topology in bilayers is elusive. Here, we fabricate bilayer structures of WTe2 films, which naturally host in-plane hyperbolic plasmons in the terahertz range. Plasmon topology is successfully modified by changing the TR and TA synergistically, manifested by the extinction spectra of unpatterned films and the polarization dependence of the plasmon intensity measured in skew ribbon arrays. Such TR- and TA-tunable topological transitions can be well explained based on the effective sheet optical conductivity by adding up those of the two films. Our study demonstrates TR as another degree of freedom for the manipulation of plasmonic topology in nanophotonics, exhibiting promising applications in bio-sensing, heat transfer and the enhancement of spontaneous emission.

physics.optics↗

arXiv2023

Urban Form and Structure Explain Variability in Spatial Inequality of Property Flood Risk among US Counties

Understanding the relationship between urban form and structure and spatial variation of property flood risk has been a longstanding challenge in urban planning and city flood risk management. Yet limited data-driven insights exist regarding the extent to which variation in spatial inequality of property flood risk in cities can be explained by heterogenous features of urban form and structure. In this study, we explore eight key features (i.e., population density, point of interest density, road density, minority segregation, income segregation, urban centrality index, gross domestic product, and human mobility index) related to urban form and structure to explain variability in spatial inequality of property flood risk among 2567 US counties. Using rich datasets related to property flood risk, we quantify spatial inequality in property flood risk and delineate features of urban form and structure using high-resolution human mobility and facility distribution data. We identify significant variation in spatial inequality of property flood risk among US counties with coastline and metropolitan counties having the greatest spatial inequality of property flood risk. The results also reveal variations in spatial inequality of property flood risk can be effectively explained based on principal components of development density, economic activity, and centrality and segregation. Using a classification and regression tree model, we demonstrate how these principal components interact and form pathways that explain levels of spatial inequality in property flood risk in US counties. The findings offer important insights for the understanding of the complex interplay between urban form and structure and spatial inequality of property flood risk and have important implications for integrated urban design strategies to address property flood risk as cities continue to expand and develop.

physics.soc-ph↗

arXiv2022

Attributed Network Embedding Model for Exposing COVID-19 Spread Trajectory Archetypes

The spread of COVID-19 revealed that transmission risk patterns are not homogenous across different cities and communities, and various heterogeneous features can influence the spread trajectories. Hence, for predictive pandemic monitoring, it is essential to explore latent heterogeneous features in cities and communities that distinguish their specific pandemic spread trajectories. To this end, this study creates a network embedding model capturing cross-county visitation networks, as well as heterogeneous features to uncover clusters of counties in the United States based on their pandemic spread transmission trajectories. We collected and computed location intelligence features from 2,787 counties from March 3 to June 29, 2020 (initial wave). Second, we constructed a human visitation network, which incorporated county features as node attributes, and visits between counties as network edges. Our attributed network embeddings approach integrates both typological characteristics of the cross-county visitation network, as well as heterogeneous features. We conducted clustering analysis on the attributed network embeddings to reveal four archetypes of spread risk trajectories corresponding to four clusters of counties. Subsequently, we identified four features as important features underlying the distinctive transmission risk patterns among the archetypes. The attributed network embedding approach and the findings identify and explain the non-homogenous pandemic risk trajectories across counties for predictive pandemic monitoring. The study also contributes to data-driven and deep learning-based approaches for pandemic analytics to complement the standard epidemiological models for policy analysis in pandemics.

cs.LG↗

arXiv2022

Quantitative Measures for Integrating Resilience into Transportation Planning Practice: Study in Texas

The objective of this study is to propose a system-level framework with quantitative measures to assess the resilience of road networks. The framework proposed in this paper can help transportation agencies incorporate resilience considerations into project development proactively and to understand the resilience performance of current road networks effectively. This study identified and implemented four quantitative metrics to classify the criticality of road segments based on critical dimensions of road network resilience, and two integrated metrics were proposed to combine all metrics to show the overall resilience performance of road segments. A case study was conducted on the Texas road networks to demonstrate the effectiveness of implementing this framework in a practical scenario. Since the data used in this study is available to other states and countries, the framework presented in this study can be adopted by other transportation agencies across the globe for regional transportation resilience assessments.

physics.soc-ph↗

arXiv2022

Characterizing Urban Lifestyle Signatures Using Motif Properties in Network of Places

The lifestyles of urban dwellers could reveal important insights regarding the dynamics and complexity of cities. Despite growing research on analysis of lifestyle patterns in cities, little is known about the characteristics of people's lifestyles patterns at urban scale. This limitation is primarily due to challenges in characterizing lifestyle patterns when human movement data is aggregated to protect the privacy of users. In this study, we model cities based on aggregated human visitation data to construct a network of places. We then examine the subgraph signatures in the networks of places to map and characterize lifestyle patterns at city scale. Location-based data from Harris County, Dallas County, New York County, and Broward County in the United States were examined to reveal lifestyle signatures in cities. For the motif analysis, two-node, three-node, and four-node motifs without location attributes were extracted from human visitation networks. Second, homogenized nodes in motifs were encoded with location categories from NAICS codes. Multiple statistical measures, including network metrics and motif properties, were quantified to characterize lifestyle signatures. The results show that: people's lifestyles in urban environments can be well depicted and quantified based on distribution and attributes of motifs in networks of places; motifs in networks of places show stability in quantity and distance as well as periodicity on weekends and weekdays indicating the stability of lifestyle patterns in cities; human visitation networks and lifestyle patterns show similarities across different metropolitan areas implying the universality of lifestyle signatures across cities. The findings provide deeper insights into urban lifestyles signatures in urban studies and provide important insights for data-informed urban planning and management.

cs.SI↗

arXiv2022

X-Pool: Cross-Modal Language-Video Attention for Text-Video Retrieval

In text-video retrieval, the objective is to learn a cross-modal similarity function between a text and a video that ranks relevant text-video pairs higher than irrelevant pairs. However, videos inherently express a much wider gamut of information than texts. Instead, texts often capture sub-regions of entire videos and are most semantically similar to certain frames within videos. Therefore, for a given text, a retrieval model should focus on the text's most semantically similar video sub-regions to make a more relevant comparison. Yet, most existing works aggregate entire videos without directly considering text. Common text-agnostic aggregations schemes include mean-pooling or self-attention over the frames, but these are likely to encode misleading visual information not described in the given text. To address this, we propose a cross-modal attention model called X-Pool that reasons between a text and the frames of a video. Our core mechanism is a scaled dot product attention for a text to attend to its most semantically similar frames. We then generate an aggregated video representation conditioned on the text's attention weights over the frames. We evaluate our method on three benchmark datasets of MSR-VTT, MSVD and LSMDC, achieving new state-of-the-art results by up to 12% in relative improvement in Recall@1. Our findings thereby highlight the importance of joint text-video reasoning to extract important visual cues according to text. Full code and demo can be found at: https://layer6ai-labs.github.io/xpool/

cs.CV↗

arXiv2021

Weakly Supervised Action Selection Learning in Video

Localizing actions in video is a core task in computer vision. The weakly supervised temporal localization problem investigates whether this task can be adequately solved with only video-level labels, significantly reducing the amount of expensive and error-prone annotation that is required. A common approach is to train a frame-level classifier where frames with the highest class probability are selected to make a video-level prediction. Frame level activations are then used for localization. However, the absence of frame-level annotations cause the classifier to impart class bias on every frame. To address this, we propose the Action Selection Learning (ASL) approach to capture the general concept of action, a property we refer to as "actionness". Under ASL, the model is trained with a novel class-agnostic task to predict which frames will be selected by the classifier. Empirically, we show that ASL outperforms leading baselines on two popular benchmarks THUMOS-14 and ActivityNet-1.2, with 10.3% and 5.7% relative improvement respectively. We further analyze the properties of ASL and demonstrate the importance of actionness. Full code for this work is available here: https://github.com/layer6ai-labs/ASL.

cs.CV↗

arXiv2019

Cross-Class Relevance Learning for Temporal Concept Localization

We present a novel Cross-Class Relevance Learning approach for the task of temporal concept localization. Most localization architectures rely on feature extraction layers followed by a classification layer which outputs class probabilities for each segment. However, in many real-world applications classes can exhibit complex relationships that are difficult to model with this architecture. In contrast, we propose to incorporate target class and class-related features as input, and learn a pairwise binary model to predict general segment to class relevance. This facilitates learning of shared information between classes, and allows for arbitrary class-specific feature engineering. We apply this approach to the 3rd YouTube-8M Video Understanding Challenge together with other leading models, and achieve first place out of over 280 teams. In this paper we describe our approach and show some empirical results.

cs.CV↗