Author name cluster

Bo Lv

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

3 papers

1 author row

AAAI Conference 2026 Conference Paper

HyCoRA: Hyper-Contrastive Role-Adaptive Learning for Role-Playing

Shihao Yang
Zhicong Lu
Yong Yang
Bo Lv
Yang Shen
Nayu Liu

Multi-character role-playing aims to equip models with the capability to simulate diverse roles. Existing methods either use one shared parameterized module across all roles or assign a separate parameterized module to each role. However, the role-shared module may ignore distinct traits of each role, weakening personality learning, while the role-specific module may overlook shared traits across multiple roles, hindering commonality modeling. In this paper, we propose a novel HyCoRA: Hyper-Contrastive Role-Adaptive learning framework, which efficiently improves multi-character role-playing agents' ability by balancing the learning of distinct and shared traits. Specifically, we propose a Hyper-Half Low-Rank Adaptation structure, where one half is a role-specific module generated by a lightweight hyper-network, and the other half is a trainable role-shared module. The role-specific module is devised to represent distinct persona signatures, while the role-shared module serves to capture common traits. Moreover, to better reflect distinct personalities across different roles, we design a hyper-contrastive learning mechanism to help the hyper-network distinguish their unique characteristics. Extensive experimental results on both English and Chinese available benchmarks demonstrate the superiority of our framework. Further GPT-4 evaluations and visual analyses also verify the capability of HyCoRA to capture role characteristics.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

SpecEM: Training-Free LLM Ensembling via Iterative Drafting, Verification, and Online Feedback

Bo Lv
Nayu Liu
Chen Tang
Xin Liu
Yue Yu
Ping Luo

Ensembles of generative large language models (LLMs) are a promising way to compensate for individual model limitations, integrating the strengths of different LLMs. Existing LLM ensemble methods, however, face limitations such as first-token delay and challenges in long-range semantic collaboration between models, Moreover, they typically assume equal voting weights for all models during ensemble, ignoring performance differences between models for a given task. In this work, we propose SpecEM, a training-free, plug-and-play LLM ensemble framework that dynamically adjusts each model's model contribution in real time based on task performance. Inspired by speculative decoding, SpecFuse iteratively performs drafting and verification, allowing models to collaborate semantically at the segment level for integrated output. Furthermore, we introduce an online feedback mechanism with multiplicative weight updates, where each model's voting weight is adjusted on-the-fly according to how often it "outperforms" others during verification stage, ensuring that stronger models exert greater influence on the ensemble during generation. Experimental results on five popular LLMs (ranging from 7B to 72B parameters) and six benchmark tasks, spanning instruction following, reasoning, commonsense, and general instruction response, demonstrate consistent performance improvements compared to state-of-the-art LLM ensemble methods.

PDF Details

JBHI Journal 2021 Journal Article

EMG Signal Filtering Based on Variational Mode Decomposition and Sub-Band Thresholding

Shihan Ma
Bo Lv
Chuang Lin
Xinjun Sheng
Xiangyang Zhu

Surface electromyography (EMG) signals are inevitably contaminated by various noise components, including powerline interference (PLI), baseline wandering (BW), and white Gaussian noise (WGN). These noises directly degrade the efficiency of EMG processing and affect the accuracy and robustness of further applications. Currently, most of the EMG filters only target one category of noise. Here, we propose a novel filter to remove all three types of noise. The noisy EMG signal is first decomposed into an ensemble of band-limited modes using variational mode decomposition (VMD). Each category of noise is located within specific modes and is separately removed in sub-bands. In particular, WGN is suppressed by soft thresholding with a noise level-dependent threshold. The denoising performance was assessed from simulated and experimental signals using three performance metrics: the root mean square error (RMSE), the improvement in signalto-noise ratio (SNRimp), and the percentage reduction in the correlation coefficient (η). Other methods, including traditional infinite impulse response (IIR) filters, empirical mode decomposition (EMD) method, and ensemble empirical mode decomposition (EEMD) method, were examined for comparison. The proposed method achieved the best performance to remove BW or WGN. It also effectively reduced PLI noise when the signal-to-noise ratio (SNR) was low. The SNR was improved by 18. 6, 19. 2, and 8. 0 dB for EMG signals corrupted with PLI, BW, and WGN at -6 dB SNR, respectively. The experimental results illustrated that noise was completely removed from resting states, and obvious spikes were distinguished from action states. For two of the ten subjects, the improved SNR reached 20 dB. This study explores the special characteristics of VMD and demonstrates the feasibility of using the VMD-based filter to denoise EMG signals. The proposed filter is efficient at removing three categories of noise and can be used for any application that requires EMG signal filtering at the preprocessing stage, such as gesture recognition and EMG decomposition.

Details DOI