arXiv subjects

Jiarui Li

Publications and source records attributed to Jiarui Li.

At least 55 records · Page 3Linked to original sources

arXiv2025

USAD: End-to-End Human Activity Recognition via Diffusion Model with Spatiotemporal Attention

The primary objective of human activity recognition (HAR) is to infer ongoing human actions from sensor data, a task that finds broad applications in health monitoring, safety protection, and sports analysis. Despite proliferating research, HAR still faces key challenges, including the scarcity of labeled samples for rare activities, insufficient extraction of high-level features, and suboptimal model performance on lightweight devices. To address these issues, this paper proposes a comprehensive optimization approach centered on multi-attention interaction mechanisms. First, an unsupervised, statistics-guided diffusion model is employed to perform data augmentation, thereby alleviating the problems of labeled data scarcity and severe class imbalance. Second, a multi-branch spatio-temporal interaction network is designed, which captures multi-scale features of sequential data through parallel residual branches with 3*3, 5*5, and 7*7 convolutional kernels. Simultaneously, temporal attention mechanisms are incorporated to identify critical time points, while spatial attention enhances inter-sensor interactions. A cross-branch feature fusion unit is further introduced to improve the overall feature representation capability. Finally, an adaptive multi-loss function fusion strategy is integrated, allowing for dynamic adjustment of loss weights and overall model optimization. Experimental results on three public datasets, WISDM, PAMAP2, and OPPORTUNITY, demonstrate that the proposed unsupervised data augmentation spatio-temporal attention diffusion network (USAD) achieves accuracies of 98.84%, 93.81%, and 80.92% respectively, significantly outperforming existing approaches. Furthermore, practical deployment on embedded devices verifies the efficiency and feasibility of the proposed method.

cs.CV↗

arXiv2025

Adversarial Robustness Unhardening via Backdoor Attacks in Federated Learning

The delicate equilibrium between user privacy and the ability to unleash the potential of distributed data is an important concern. Federated learning, which enables the training of collaborative models without sharing of data, has emerged as a privacy-centric solution. This approach brings forth security challenges, notably poisoning and backdoor attacks where malicious entities inject corrupted data into the training process, as well as evasion attacks that aim to induce misclassifications at test time. Our research investigates the intersection of adversarial training, a common defense method against evasion attacks, and backdoor attacks within federated learning. We introduce Adversarial Robustness Unhardening (ARU), which is employed by a subset of adversarial clients to intentionally undermine model robustness during federated training, rendering models susceptible to a broader range of evasion attacks. We present extensive experiments evaluating ARU's impact on adversarial training and existing robust aggregation defenses against poisoning and backdoor attacks. Our results show that ARU can substantially undermine adversarial training's ability to harden models against test-time evasion attacks, and that adversaries employing ARU can even evade robust aggregation defenses that often neutralize poisoning or backdoor attacks.

cs.LG↗

arXiv2025

Multi-Agent Path Finding via Finite-Horizon Hierarchical Factorization

We present a novel algorithm for large-scale Multi-Agent Path Finding (MAPF) that enables fast, scalable planning in dynamic environments such as automated warehouses. Our approach introduces finite-horizon hierarchical factorization, a framework that plans one step at a time in a receding-horizon fashion. Robots first compute individual plans in parallel, and then dynamically group based on spatio-temporal conflicts and reachability. The framework accounts for conflict resolution, and for immediate execution and concurrent planning, significantly reducing response time compared to offline algorithms. Experimental results on benchmark maps demonstrate that our method achieves up to 60% reduction in time-to-first-action while consistently delivering high-quality solutions, outperforming state-of-the-art offline baselines across a range of problem sizes and planning horizons.

cs.RO↗

arXiv2025

Electronic structure of compressively strained thin film La$_2$PrNi$_2$O$_7$

The discovery of superconductivity in the bulk nickelates under high pressure is a major advance in physics. The recent observation of superconductivity at ambient pressure in compressively strained bilayer nickelate thin films has now enabled direct characterization of the superconducting phase through angle resolved photoemission spectroscopy (ARPES). Here we present an in-situ ARPES study of compressively strained La$_2$PrNi$_2$O$_7$ films grown by oxide molecular beam epitaxy, and the ozone treated counterparts with an onset T$_c$ of 40 K, supplemented with results from pulsed laser deposition films with similar T$_c$. We resolve a systematic strain-driven electronic band shift with respect to that of bulk crystals, in qualitative agreement with density functional theory (DFT) calculations. However, the strongly renormalized flat 3$d_{z2}$ band shifts a factor of 5-10 smaller than anticipated by DFT. Furthermore, it stays ~70 meV below the Fermi level, contradicting the expectation that superconductivity results from the high density of states of this band at the Fermi level. We also observed a non-trivial k$_z$ dispersion of the cuprate-like 3$d_{x2-y2}$ band. Combined with results from both X-ray diffraction and DFT, we suggest that the strained films are under ~5 GPa effective pressure, considerably larger than the naïve expectation from the DFT relaxed structure. Finally, the ~70 meV energy position is intriguingly close to the collective mode coupling more prominently seen in thin films, in the energy range of both oxygen related phonons and the maximum of the spin excitation spectrum.

cond-mat.supr-con↗

arXiv2025

Process Optimization and Deployment for Sensor-Based Human Activity Recognition Based on Deep Learning

Sensor-based human activity recognition is a key technology for many human-centered intelligent applications. However, this research is still in its infancy and faces many unresolved challenges. To address these, we propose a comprehensive optimization process approach centered on multi-attention interaction. We first utilize unsupervised statistical feature-guided diffusion models for highly adaptive data enhancement, and introduce a novel network architecture-Multi-branch Spatiotemporal Interaction Network, which uses multi-branch features at different levels to effectively Sequential ), which uses multi-branch features at different levels to effectively Sequential spatio-temporal interaction to enhance the ability to mine advanced latent features. In addition, we adopt a multi-loss function fusion strategy in the training phase to dynamically adjust the fusion weights between batches to optimize the training results. Finally, we also conducted actual deployment on embedded devices to extensively test the practical feasibility of the proposed method in existing work. We conduct extensive testing on three public datasets, including ablation studies, comparisons of related work, and embedded deployments.

eess.SP↗

arXiv2025

Doping dependence of 2-spinon excitations in the doped 1D cuprate Ba$_2$CuO$_{3+δ}$

Recent photoemission experiments on the quasi-one-dimensional Ba-based cuprates suggest that doped holes experience an attractive potential not captured using the simple Hubbard model. This observation has garnered significant attention due to its potential relevance to Cooper pair formation in high-$T_c$ cuprate superconductors. To scrutinize this assertion, we examined signatures of such an attractive potential in doped 1D cuprates Ba$_2$CuO$_{3+δ}$ by measuring the dispersion of the 2-spinon excitations using Cu $L_3$-edge resonant inelastic X-ray scattering (RIXS). Upon doping, the 2-spinon excitations appear to weaken, with a shift of the minimal position corresponding to the nesting vector of the Fermi points, $q_F$. Notably, we find that the energy scale of the 2-spinons near the Brillouin zone boundary is substantially softened compared to that predicted by the Hubbard model in one-dimension. Such a discrepancy implies missing ingredients, which lends support for the presence of an additional attractive potential between holes.

cond-mat.str-el↗

arXiv2025

VicKAM: Visual Conceptual Knowledge Guided Action Map for Weakly Supervised Group Activity Recognition

Existing weakly supervised group activity recognition methods rely on object detectors or attention mechanisms to capture key areas automatically. However, they overlook the semantic information associated with captured areas, which may adversely affect the recognition performance. In this paper, we propose a novel framework named Visual Conceptual Knowledge Guided Action Map (VicKAM) which effectively captures the locations of individual actions and integrates them with action semantics for weakly supervised group activity recognition.It generates individual action prototypes from training set as visual conceptual knowledge to bridge action semantics and visual representations. Guided by this knowledge, VicKAM produces action maps that indicate the likelihood of each action occurring at various locations, based on image correlation theorem. It further augments individual action maps using group activity related statistical information, representing individual action distribution under different group activities, to establish connections between action maps and specific group activities. The augmented action map is incorporated with action semantic representations for group activity recognition.Extensive experiments on two public benchmarks, the Volleyball and the NBA datasets, demonstrate the effectiveness of our proposed method, even in cases of limited training data. The code will be released later.

cs.CV↗

arXiv2025

The Classical-to-Quantum Crossover in strain-induced ferroelectric transition in SrTiO$_3$ membranes

Mechanical strain presents an effective control over symmetry-breaking phase transitions. In quantum paralelectric SrTiO3, strain can induce the ferroelectric transition via modification of local Ti potential landscape. However, brittle bulk materials can only withstand limited strain range (~0.1%). Taking advantage of nanoscopically-thin freestanding membranes, we demonstrated in-situ strain-induced reversible ferroelectric transition in a single freestanding SrTiO3 membranes. We measure the ferroelectric order by detecting the local anisotropy of the Ti 3d orbital using X-ray linear dichroism at the Ti-K pre-edge, while the strain is determined by X-ray diffraction. With reduced thickness, the SrTiO3 membranes remain elastic with >1% tensile strain cycles. A robust displacive ferroelectricity appears beyond a temperature-dependent critical strain. Interestingly, we discover a crossover from a classical ferroelectric transition to a quantum regime at low temperatures, which enhances strain-induced ferroelectricity. Our results offer a new opportunities to strain engineer functional properties in low dimensional quantum materials and provide new insights into the role of the ferroelectric fluctuations in quantum paraelectric SrTiO3.

cond-mat.mtrl-sci↗

arXiv2025

Large moiré superstructure of stacked incommensurate charge density waves

Recent advances in van der Waals heterostructures have opened the new frontier of moiré physics, whereby tuning the interlayer twist angle or adjusting lattice parameter mismatch have led to a plethora of exotic phenomena such as unconventional superconductivity and fractional quantum spin Hall effect. We extend the concept of moiré engineering to materials that host incommensurate orders, where we discovered a long-period, thermally-hysteretic moiré superlattice in a layered charge density wave (CDW) compound, EuTe$_\text{4}$. Using high-momentum-resolution X-ray diffraction performed on ultrathin flakes, we found two coexisting, incommensurate CDWs with slightly mismatched in-plane wavevectors. The interaction between these two CDWs leads to their joint commensuration with the high-symmetry lattice as well as a large moiré superstructure with an in-plane period of 13.6~nm. Due to different out-of-plane orders of the incommensurate CDWs, the moiré superstructure exhibits a clear thermal hysteresis, accounting for the large hysteresis observed in electrical resistivity and numerous metastable states induced by light or electrical pulses. Our findings pave the way for a new development in moiré engineering based on an incommensurate lattice. They further highlight the important role of interlayer ordering in determining the macroscopic properties of these stacked incommensurate structures.

cond-mat.str-el↗

arXiv2024

Prototype-Driven Multi-Feature Generation for Visible-Infrared Person Re-identification

The primary challenges in visible-infrared person re-identification arise from the differences between visible (vis) and infrared (ir) images, including inter-modal and intra-modal variations. These challenges are further complicated by varying viewpoints and irregular movements. Existing methods often rely on horizontal partitioning to align part-level features, which can introduce inaccuracies and have limited effectiveness in reducing modality discrepancies. In this paper, we propose a novel Prototype-Driven Multi-feature generation framework (PDM) aimed at mitigating cross-modal discrepancies by constructing diversified features and mining latent semantically similar features for modal alignment. PDM comprises two key components: Multi-Feature Generation Module (MFGM) and Prototype Learning Module (PLM). The MFGM generates diversity features closely distributed from modality-shared features to represent pedestrians. Additionally, the PLM utilizes learnable prototypes to excavate latent semantic similarities among local features between visible and infrared modalities, thereby facilitating cross-modal instance-level alignment. We introduce the cosine heterogeneity loss to enhance prototype diversity for extracting rich local features. Extensive experiments conducted on the SYSU-MM01 and LLCM datasets demonstrate that our approach achieves state-of-the-art performance. Our codes are available at https://github.com/mmunhappy/ICASSP2025-PDM.

cs.CV↗

arXiv2024

Signatures of polarized chiral spin disproportionation in rare earth nickelates

In rare earth nickelates (RENiO$_3$), electron-lattice coupling drives a concurrent metal-to-insulator and bond disproportionation phase transition whose microscopic origin has long been the subject of active debate. Of several proposed mechanisms, here we test the hypothesis that pairs of self-doped ligand holes spatially condense to provide local spin moments that are antiferromagnetically coupled to Ni spins. These singlet-like states provide a basis for long-range bond and spiral spin order. Using magnetic resonant X-ray scattering on NdNiO$_3$ thin films, we observe the chiral nature of the spin-disproportionated state, with spin spirals propagating along the crystallographic (101)$_\mathrm{ortho}$ direction. These spin spirals are found to preferentially couple to X-ray helicity, establishing the presence of a hitherto-unobserved macroscopic chirality. The presence of this chiral magnetic configuration suggests a potential multiferroic coupling between the noncollinear magnetic arrangement and improper ferroelectric behavior as observed in prior studies on NdNiO$_3$ (101)$_\mathrm{ortho}$ films and RENiO$_3$ single crystals. Experimentally constrained theoretical double-cluster calculations confirm the presence of an energetically stable spin-disproportionated state with Zhang-Rice singlet-like combinations of Ni and ligand moments.

cond-mat.str-el↗

arXiv2024

Electronically Amplified Electron-Phonon Interaction and Metal-Insulator Transition in Perovskite Nickelates

The relative role of electron-electron and electron-lattice interactions in driving the metal-insulator transition in perovskite nickelates opens a rare window into the non-trivial interplay of the two important degrees of freedom in solids. The most promising solution is to extract the electronic and lattice contributions during the phase transition by performing high-resolution spectroscopy measurements. Here, we present a three-dimensional electronic structure study of Nd1-xSrxNiO3 (x = 0 and 0.175) thin films with unprecedented accuracy, in which the low energy fermiology has a quantitative agreement with model simulations and first-principles calculations. Two characteristic phonons, the octahedral rotational and breathing modes, are illustrated to be coupled with the electron dynamics in the metallic phase, showing a kink structure along the band dispersion, as well as a hump feature in the energy spectrum. Entering the insulating state, the electron-phonon interaction is amplified by strong electron correlations, transforming the mobile large polarons at high temperatures to localized small polarons in the ground state. Moreover, the analysis of quasiparticle residue enables us to establish a transport-spectroscopy correspondence in Nd1-xSrxNiO3 thin films. Our findings demonstrate the essential role of electron-lattice interaction enhanced by the electronic correlation to stabilize the insulating phase in the perovskite nickelates.

cond-mat.str-el↗

arXiv2024

Coexistence of interacting charge density waves in a layered semiconductor

Coexisting orders are key features of strongly correlated materials and underlie many intriguing phenomena from unconventional superconductivity to topological orders. Here, we report the coexistence of two interacting charge-density-wave (CDW) orders in EuTe4, a layered crystal that has drawn considerable attention owing to its anomalous thermal hysteresis and a semiconducting CDW state despite the absence of perfect FS nesting. By accessing unoccupied conduction bands with time- and angle-resolved photoemission measurements, we find that mono- and bi-layers of Te in the unit cell host different CDWs that are associated with distinct energy gaps. The two gaps display dichotomous evolutions following photoexcitation, where the larger bilayer CDW gap exhibits less renormalization and faster recovery. Surprisingly, the CDW in the Te monolayer displays an additional momentum-dependent gap renormalization that cannot be captured by density-functional theory calculations. This phenomenon is attributed to interlayer interactions between the two CDW orders, which account for the semiconducting nature of the equilibrium state. Our findings not only offer microscopic insights into the correlated ground state of EuTe4 but also provide a general non-equilibrium approach to understand coexisting, layer-dependent orders in a complex system.

cond-mat.str-el↗

arXiv2024

Enhancing LLM Factual Accuracy with RAG to Counter Hallucinations: A Case Study on Domain-Specific Queries in Private Knowledge-Bases

We proposed an end-to-end system design towards utilizing Retrieval Augmented Generation (RAG) to improve the factual accuracy of Large Language Models (LLMs) for domain-specific and time-sensitive queries related to private knowledge-bases. Our system integrates RAG pipeline with upstream datasets processing and downstream performance evaluation. Addressing the challenge of LLM hallucinations, we finetune models with a curated dataset which originates from CMU's extensive resources and annotated with the teacher model. Our experiments demonstrate the system's effectiveness in generating more accurate answers to domain-specific and time-sensitive inquiries. The results also revealed the limitations of fine-tuning LLMs with small-scale and skewed datasets. This research highlights the potential of RAG systems in augmenting LLMs with external datasets for improved performance in knowledge-intensive tasks. Our code and models are available on Github.

cs.CL↗

arXiv2024

Highly confined epsilon-near-zero- and surface-phonon polaritons in SrTiO3 membranes

Recent theoretical studies have suggested that transition metal perovskite oxide membranes can enable surface phonon polaritons in the infrared range with low loss and much stronger subwavelength confinement than bulk crystals. Such modes, however, have not been experimentally observed so far. Here, using a combination of far-field Fourier-transform infrared (FTIR) spectroscopy and near-field synchrotron infrared nanospectroscopy (SINS) imaging, we study the phonon-polaritons in a 100 nm thick freestanding crystalline membrane of SrTiO3 transferred on metallic and dielectric substrates. We observe a symmetric-antisymmetric mode splitting giving rise to epsilon-near-zero and Berreman modes as well as highly confined (by a factor of 10) propagating phonon polaritons, both of which result from the deep-subwavelength thickness of the membranes. Theoretical modeling based on the analytical finite-dipole model and numerical finite-difference methods fully corroborate the experimental results. Our work reveals the potential of oxide membranes as a promising platform for infrared photonics and polaritonics.

cond-mat.mtrl-sci↗

arXiv2024

Two Distinct Charge Orders in Infinite-layer PrNiO2+δ revealed by Resonant X-ray Diffraction

A broken translation symmetry has recently been revealed in infinite-layer nickelates, which has piqued considerable interest in its origin and relation to superconductivity, as well as in its comparison to charge order in cuprates. Here, by performing resonant x-ray scattering measurements in thin films of infinite-layer PrNiO2+δ, we find that the superlattice reflection at the Ni L3 absorption edge differs considerably from that at the Pr M5 resonance in their dependence on energy, temperature, and local symmetry, indicating they are two distinct charge orders despite the same in-plane wavevectors. These dissimilarities might be related to the excess oxygen dopants, considering that the resonant reflections were observed in an incompletely reduced PrNiO2+δ film. In addition, our azimuthal analysis suggests that the oxygen ligands should play a pivotal role in the charge modulation revealed at the Ni L3 resonance.

cond-mat.supr-con↗

arXiv2023

High-speed impulsive stimulated Brillouin microscopy

Brillouin microscopy, which maps elastic modulus from the frequency shift of scattered light, has evolved to a faster speed for the investigation of rapid biomechanical changes. Impulsive stimulated Brillouin scattering (ISBS) spectroscopy has the potential to speed up measurement through the resonant amplification interaction from pulsed excitation and time-domain continuous detection. However, significant progress has not been achieved due to the limitation in signal-to-noise ratio (SNR) and the corresponding need for excessive averaging to maintain high spectral precision. Moreover, the limited spatial resolution also hinders its application in mechanical imaging. Here, by scrutinizing the SNR model, we design a high-speed ISBS microscope through multi-parameter optimization including phase, reference power, and acquisition time. Leveraging this, with the further assistance of the Matrix Pencil method for data processing, three-dimensional mechanical images are mapped under multiple contrast mechanisms for a millimeter-scale polydimethylsiloxane pattern immersed in methanol, enabling the identification of these two transparent materials without any contact or labeling. Our experimental results demonstrate the capability to maintain high spectral precision and resolution at a sub-millisecond integration time for one pixel. With a two-order improvement in the speed and a tenfold improvement in the spatial resolution over the state-of-the-art systems, this method makes it possible for ISBS microscopes to sensitively investigate rapid mechanical changes in time and space.

physics.optics↗

arXiv2023

Bipartite Graph Pre-training for Unsupervised Extractive Summarization with Graph Convolutional Auto-Encoders

Pre-trained sentence representations are crucial for identifying significant sentences in unsupervised document extractive summarization. However, the traditional two-step paradigm of pre-training and sentence-ranking, creates a gap due to differing optimization objectives. To address this issue, we argue that utilizing pre-trained embeddings derived from a process specifically designed to optimize cohensive and distinctive sentence representations helps rank significant sentences. To do so, we propose a novel graph pre-training auto-encoder to obtain sentence embeddings by explicitly modelling intra-sentential distinctive features and inter-sentential cohesive features through sentence-word bipartite graphs. These pre-trained sentence representations are then utilized in a graph-based ranking algorithm for unsupervised summarization. Our method produces predominant performance for unsupervised summarization frameworks by providing summary-worthy sentence representations. It surpasses heavy BERT- or RoBERTa-based sentence representations in downstream tasks.

cs.CL↗