arXiv subjects

Defeng

Publications and source records attributed to Defeng.

7 recordsLinked to original sources

arXiv2025

Pseudo Strong Labels from Frame-Level Predictions for Weakly Supervised Sound Event Detection

Weakly Supervised Sound Event Detection (WSSED), which relies on audio tags without precise onset and offset times, has become prevalent due to the scarcity of strongly labeled data that includes exact temporal boundaries for events. This study introduces Frame-level Pseudo Strong Labeling (FPSL) to overcome the lack of temporal information in WSSED by generating pseudo strong labels from frame-level predictions. This enhances temporal localization during training and addresses the limitations of clip-wise weak supervision. We validate our approach across three benchmark datasets (DCASE2017 Task 4, DCASE2018 Task 4, and UrbanSED) and demonstrate significant improvements in key metrics such as the Polyphonic Sound Detection Scores (PSDS), event-based F1 scores, and intersection-based F1 scores. For example, Convolutional Recurrent Neural Networks (CRNNs) trained with FPSL outperform baseline models by 4.9% in PSDS1 on DCASE2017, 7.6% on DCASE2018, and 1.8% on UrbanSED, confirming the effectiveness of our method in enhancing model performance.

eess.AS↗

arXiv2024

Impact of Noisy Labels on Sound Event Detection: Deletion Errors Are More Detrimental Than Insertion Errors

This study explores the critical but underexamined impact of label noise on Sound Event Detection (SED), which requires both sound identification and precise temporal localization. We categorize label noise into deletion, insertion, substitution, and subjective types and systematically evaluate their effects on SED using synthetic and real-life datasets. Our analysis shows that deletion noise significantly degrades performance, while insertion noise is relatively benign. Moreover, loss functions effective against classification noise do not perform well for SED due to intra-class imbalance between foreground sound events and background sounds. We demonstrate that loss functions designed to address data imbalance in SED can effectively reduce the impact of noisy labels on system performance. For instance, halving the weight of background sounds in a synthetic dataset improved macro-F1 and micro-F1 scores by approximately $9\%$ with minimal Error Rate increase, with consistent results in real-life datasets. This research highlights the nuanced effects of noisy labels on SED systems and provides practical strategies to enhance model robustness, which are pivotal for both constructing new SED datasets and improving model performance, including efficient utilization of soft and crowdsourced labels.

eess.AS↗

arXiv2021

Unitary Approximate Message Passing for Sparse Bayesian Learning

Sparse Bayesian learning (SBL) can be implemented with low complexity based on the approximate message passing (AMP) algorithm. However, it does not work well for a generic measurement matrix, which may cause AMP to diverge. Damped AMP has been used for SBL to alleviate the problem at the cost of reducing convergence speed. In this work, we propose a new SBL algorithm based on structured variational inference, leveraging AMP with a unitary transformation (UAMP). Both single measurement vector and multiple measurement vector problems are investigated. It is shown that, compared to state-of-the-art AMP-based SBL algorithms, the proposed UAMP-SBL is more robust and efficient, leading to remarkably better performance.

eess.SP↗

arXiv2020

Simultaneous measurements on the electron and X-ray spectra from laser-irradiated near-critical-density double-layer targets at relativistic intensity

We report the experimental results of simultaneous measurements on the electron and X-ray spectra from near-critical-density (NCD) double-layer targets irradiated by relativistic femtosecond pulses at the intensity of 5E19 W/cm^2. The dependence of the electron and X-ray spectra on the density and thickness of the NCD layer was studied. For the optimal targets, electrons with temperature of 5.5 MeV and X-rays with critical energy of 5 keV were obtained. 2D particle-in-cell simulations based on the experimental parameters confirm the electrons are accelerated in the plasma channel through direct laser acceleration, resulting in temperature significantly higher than the pondermotive temperature. Bright X-rays are generated from betatron emission and Thomson backscattering before the electrons leave the double-layer targets.

physics.plasm-ph↗

arXiv2016

Approximate Message Passing with Nearest Neighbor Sparsity Pattern Learning

We consider the problem of recovering clustered sparse signals with no prior knowledge of the sparsity pattern. Beyond simple sparsity, signals of interest often exhibits an underlying sparsity pattern which, if leveraged, can improve the reconstruction performance. However, the sparsity pattern is usually unknown a priori. Inspired by the idea of k-nearest neighbor (k-NN) algorithm, we propose an efficient algorithm termed approximate message passing with nearest neighbor sparsity pattern learning (AMP-NNSPL), which learns the sparsity pattern adaptively. AMP-NNSPL specifies a flexible spike and slab prior on the unknown signal and, after each AMP iteration, sets the sparse ratios as the average of the nearest neighbor estimates via expectation maximization (EM). Experimental results on both synthetic and real data demonstrate the superiority of our proposed algorithm both in terms of reconstruction performance and computational complexity.

cs.IT↗

arXiv2015

Message-Passing Receiver for Joint Channel Estimation and Decoding in 3D Massive MIMO-OFDM Systems

In this paper, we address the message-passing receiver design for the 3D massive MIMO-OFDM systems. With the aid of the central limit argument and Taylor-series approximation, a computationally efficient receiver that performs joint channel estimation and decoding is devised by the framework of expectation propagation. Specially, the local belief defined at the channel transition function is expanded up to the second order with Wirtinger calculus, to transform the messages sent by the channel transition function to a tractable form. As a result, the channel impulse response (CIR) between each pair of antennas is estimated by Gaussian message passing. In addition, a variational expectation-maximization (EM)-based method is derived to learn the channel power-delay-profile (PDP). The proposed joint algorithm is assessed in 3D massive MIMO systems with spatially correlated channels, and the empirical results corroborate its superiority in terms of performance and complexity.

cs.IT↗

arXiv2012

Accurate Sampling Timing Acquisition for Baseband OFDM Power-line Communication in Non-Gaussian Noise

In this paper, a novel technique is proposed to address the joint sampling timing acquisition for baseband and broadband power-line communication (BB-PLC) systems using Orthogonal-Frequency-Division-Multiplexing (OFDM), including the sampling phase offset (SPO) and the sampling clock offset (SCO). Under pairwise correlation and joint Gaussian assumption of received signals in frequency domain, an approximated form of the log-likelihood function is derived. Instead of a high complexity two-dimension grid-search on the likelihood function, a five-step method is employed for accurate estimations. Several variants are presented in the same framework with different complexities. Unlike conventional pilot-assisted schemes using the extra phase rotations within one OFDM block, the proposed technique turns to the phase rotations between adjacent OFDM blocks. Analytical expressions of the variances and biases are derived. Extensive simulation results indicate significant performance improvements over conventional schemes. Additionally, effects of several noise models including non-Gaussianity, cyclo-stationarity, and temporal correlation are analyzed and simulated. Robustness of the proposed technique against violation of the joint Gaussian assumption is also verified by simulations.

cs.IT↗