26 July 2023 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
Differential beamforming techniques have gained significant attention due to their frequency-independent beampatterns, applicability for small apertures, and super-directivity. It is commonly known that the main lobe of a beam pattern based on a linear array is typically aligned with the end-fire direction. However, in some application scenarios, aligning the main lobe in the end-fire direction may not meet the actual requirements. This article focuses on studying the steering problem of differential beamformers based on linear acoustics vector sensor arrays (LAVSs). We propose a steerable differential beamformer for LAVSs, which addresses the problem of non-steerable beam patterns of linear arrays. Our approach involves approximating the beam pattern using a series expansion of the weighted steering vector to achieve frequency-independent differential directivity. The major contributions of this article include, but are not limited to, the following: First, we derive the first-order to third-order differential directivity based on the acoustics vector sensor (AVS) signal model, and obtain the weights of the corresponding channels using the least squares method. Second, we provide a theoretical analysis of the steering differential beamformer, demonstrating that the maximum directivity factor (DF) is a function of the steering angle. Finally, we conduct simulation experiments to validate the proposed method.
15 July 2024 in IEEE Transactions on Antennas and Propagation
A co-design approach is proposed in this communication for an efficient design of frequency scanning coupled-patch arrays (CPAs) with wide scanning range and high scanning rate. It combines characteristic mode analysis (CMA) and coupling matrix analysis, where the former is employed for radiation pattern synthesis while the latter tunes the feeding reflection and mode excitation based on a coupled-resonator network model. With this systematic approach, it becomes efficient to control both beam scanning properties and feeding reflection. 3 rd -order and 6 th -order design examples are provided to validate the proposed method. The 6 th -order CPA achieves a wide scanning range of 73° (from -45° to 28° across broadside) and a high scanning rate of 26.1°/1%. Both are higher than other reported CPAs.
16 March 2023 in Journal of Applied Physics
It has been demonstrated that sound waves carrying orbital angular momentum undergo frequency modulation after being reflected by a rotating object. In this paper, we, observed the rotational Doppler shift phenomenon of the acoustic vortex in an underwater environment by using a Fresnel-spiral zone plate for the construction of the focused acoustic vortex with controllable topological charge. The positive and negative rotation speeds of the spinning object are precisely determined with the rotational frequency shift and the acoustic vortex's topological charge. The rotational Doppler effect of acoustic vortex provides a possibility to improve the detection and recognition accuracy of Autonomous Underwater Vehicles with propellers.
21 June 2024 in Physical Communication
30 June 2025 in IEEE Sensors Journal
A novel underwater noise learning model with a local-global convolution-enhanced Transformer(NL-LGCET) is proposed to underwater acoustic signal denoising within complex marine environments.The model introduces a Local-Global Convolution-Enhanced Transformer (LGCET) with a dual-branch mechanism for parallel extraction of local and global temporal features, surpassing traditional serial structures. A Convolution-Enhanced Transformer (CET) block integrates multi-head self-attention with an advanced depthwise separable convolution to capture intricate time-frequency features. Replacing the traditional STFT, a learnable one-dimensional convolution module preserves phase characteristics and adapts to environmental noise for precise feature extraction. Additionally, a noise learning module and time-domain residual mechanism ensure accurate noise estimation and signal reconstruction. Experimental results shows that the proposed model achieves 19.11 dB scale-invariant signal-to-noise ratio (SISNR) and 20.40 dB signal-to-distortion ratio (SDR) on the ShipsEar dataset, representing significant improvements over the optimal baseline. Furthermore, the model exhibits robust generalization capabilities in unknown ship scenarios and real scenarios.
11 April 2025 in Proceedings of the AAAI Conference on Artificial Intelligence
Recent advancements in AI alignment techniques have significantly improved the alignment of large language models (LLMs) with static human preferences. However, the dynamic nature of human preferences can render some prior training data outdated or even erroneous, ultimately causing LLMs to deviate from contemporary human preferences and societal norms. Existing methodologies, either curation of new data for continual alignment or manual correction of outdated data for re-alignment, demand costly human resources. To address this, we propose a novel approach, LLM BehAvior Correction with INfluence FunCtion REcall and Post-Training (LANCET), which needs no human involvement. LANCET consists of two phases: (1) using a new method LinFAC to efficiently identify the training data that significantly impact undesirable model outputs, and (2) applying an novel Influence-driven Bregman Optimization (IBO) technique to adjust the model’s outputs based on these influence distributions. Our experiments show that LANCET effectively and efficiently corrects inappropriate behaviors of LLMs while preserving model utility. Further more, LANCET exhibits stronger generalization ability than all baselines under out-of-distribution harmful prompts, offering better interpretability and compatibility with real-world applications of LLMs.