Lu Ma

Jingyu Zhou, Lu Ma, Hao Liang et al.

Recent advances in large language models (LLMs) have shown that reasoning ability can be significantly enhanced through Reinforcement Learning with Verifiable Rewards (RLVR). Group Relative Policy Optimization (GRPO) has emerged as the de facto approach for RLVR, inspiring numerous variants. However, our mathematical analysis reveals that these methods are fundamentally weighted variations of GRPO. We provide a unified view, demonstrating that their reliance on static or overly simplistic weighting schemes tied to sample difficulty prevents adaptation to a model's evolving capabilities. This creates a significant loss scale issue, where training disproportionately focuses on certain difficulty levels at the expense of others, hindering overall performance. To address these limitations, we introduce \textbf{Difficulty-Aware Reweighting Policy Optimization (DARO)}, a method that dynamically adjusts the loss contribution of each difficulty group based on the model's learning state. Extensive experiments on Qwen2.5-Math-1.5B, Qwen2.5-Math-7B, and Llama3.1-8B show that DARO outperforms four leading baselines across six math benchmarks, achieving significantly faster convergence and superior final performance.

Baolu Li, Hongkai Yu, Huiming Sun et al.

The synchrotron light source, a cutting-edge large-scale user facility, requires autonomous synchrotron beamline operations, a crucial technique that should enable experiments to be conducted automatically, reliably, and safely with minimum human intervention. However, current state-of-the-art synchrotron beamlines still heavily rely on human safety oversight. To bridge the gap between automated and autonomous operation, a computer vision-based system is proposed, integrating deep learning and multiview cameras for real-time collision detection. The system utilizes equipment segmentation, tracking, and geometric analysis to assess potential collisions with transfer learning that enhances robustness. In addition, an interactive annotation module has been developed to improve the adaptability to new object classes. Experiments on a real beamline dataset demonstrate high accuracy, real-time performance, and strong potential for autonomous synchrotron beamline operations.

Lu Ma, Song Yang, Yaguang Gong et al.

Acoustic Echo Cancellation (AEC) plays a key role in speech interaction by suppressing the echo received at microphone introduced by acoustic reverberations from loudspeakers. Since the performance of linear adaptive filter (AF) would degrade severely due to nonlinear distortions, background noises, and microphone clipping in real scenarios, deep learning has been employed for AEC for its good nonlinear modelling ability. In this paper, we constructed an end-to-end multi-scale attention neural network for AEC. Temporal convolution is first used to transform waveform into spectrogram. The spectrograms of the far-end reference and the near-end mixture are concatenated, and fed to a temporal convolution network (TCN) with stacked dilated convolution layers. Attention mechanism is performed among these representations from different layers to adaptively extract relevant features by referring to the previous hidden state in the encoder long short-term memory (LSTM) unit. The representations are weighted averaged and fed to the encoder LSTM for the near-end speech estimation. Experiments show the superiority of our method in terms of the echo return loss enhancement (ERLE) for single-talk periods and the perceptual evaluation of speech quality (PESQ) score for double-talk periods in background noise and nonlinear distortion scenarios.

Lu Ma, Song Yang, Yaguang Gong et al.

This paper proposes an noise type classification aided attention-based neural network approach for monaural speech enhancement. The network is constructed based on a previous work by introducing a noise classification subnetwork into the structure and taking the classification embedding into the attention mechanism for guiding the network to make better feature extraction. Specifically, to make the network an end-to-end way, an audio encoder and decoder constructed by temporal convolution is used to make transformation between waveform and spectrogram. Additionally, our model is composed of two long short term memory (LSTM) based encoders, two attention mechanism, a noise classifier and a speech mask generator. Experiments show that, compared with OM-LSA and the previous work, the proposed noise classification aided attention-based approach can achieve better performance in terms of speech quality (PESQ). More promisingly, our approach has better generalization ability to unseen noise conditions.

Lu Ma, Xintian Wang, Song Yang et al.

Teaching with the cooperation of expert teacher and assistant teacher, which is the so-called "double-teachers classroom", i.e., the course is giving by the expert online and presented through projection screen at the classroom, and the teacher at the classroom performs as an assistant for guiding the students in learning, is becoming more prevalent in today's teaching method for K-12 education. For monitoring the teaching quality, a microphone clipped on the assistant's neckline is always used for voice recording, then fed to the downstream tasks of automatic speech recognition (ASR) and neural language processing (NLP). However, besides its voice, there would be some other interfering voices, including the expert's one and the student's one. Here, we propose to extract the assistant' voices from the perspective of sound event detection, i.e., the voices are classified into four categories, namely the expert, the teacher, the mixture of them, and the background. To make frame-level identification, which is important for grabbing sensitive words for the downstream tasks, a multi-scale temporal convolution neural network is constructed with stacked dilated convolutions for considering both local and global properties. These features are concatenated and fed to a classification network constructed by three linear layers. The framework is evaluated on simulated data and real-world recordings, giving considerable performance in terms of precision and recall, compared with some classical classification methods.

Lu Ma, Song Yang, Yaguang Gong et al.

Acoustic Echo Cancellation (AEC) whose aim is to suppress the echo originated from acoustic coupling between loudspeakers and microphones, plays a key role in voice interaction. Linear adaptive filter (AF) is always used for handling this problem. However, since there would be some severe effects in real scenarios, such nonlinear distortions, background noises, and microphone clipping, it would lead to considerable residual echo, giving poor performance in practice. In this paper, we propose an end-to-end network structure for echo cancellation, which is directly done on time-domain audio waveform. It is transformed to deep representation by temporal convolution, and modelled by Long Short-Term Memory (LSTM) for considering temporal property. Since time delay and severe reverberation may exist at the near-end with respect to the far-end, a local attention is employed for alignment. The network is trained using multitask learning by employing an auxiliary classification network for double-talk detection. Experiments show the superiority of our proposed method in terms of the echo return loss enhancement (ERLE) for single-talk periods and the perceptual evaluation of speech quality (PESQ) score for double-talk periods in background noise and nonlinear distortion scenarios.

Lu Ma, Haiping Zhang, Pei Zhao et al.

Wakeup is the primary function in voice interaction which is the mainstream scheme in man-machine interaction (HMI) applications for smart home. All devices will response if the same wake-up word is used for all devices. This will bring chaos and reduce user quality of experience (QoE). The only way to solve this problem is to make all the devices in the same wireless local area network (WLAN) competing to wake-up based on the same scoring rule. The one closest to the user would be selected for response. To this end, a competitive wakeup scheme is proposed in this paper with elaborately designed calibration method for receiving energy of microphones. Moreover, the user orientation is assisted to determine the optimal device. Experiments reveal the feasibility and validity of this scheme.

Lu Ma, Xin Zhao, Pei Zhao et al.

Since space-domain information can be utilized, microphone array beamforming is often used to enhance the quality of the speech by suppressing directional disturbance. However, with the increasing number of microphone, the complexity would be increased. In this paper, a concise beamforming scheme using Maximum Signal-to-Noise Ratio (SNR) filter is proposed to reduce the beamforming complexity. The maximum SNR filter is implemented by using the estimated direction-of-arrival (DOA) of the speech source localization (SSL) and the solving method of independent vector analysis (IVA). Our experiments show that when compared with other widely-used algorithms, the proposed algorithm obtain higher gain of signal-to-interference and noise ratio (SINR).

Lu Ma, Hua Huang, Pei Zhao et al.

Acoustic Echo Cancellation (AEC) plays a key role in voice interaction. Due to the explicit mathematical principle and intelligent nature to accommodate conditions, adaptive filters with different types of implementations are always used for AEC, giving considerable performance. However, there would be some kinds of residual echo in the results, including linear residue introduced by mismatching between estimation and the reality and non-linear residue mostly caused by non-linear components on the audio devices. The linear residue can be reduced with elaborate structure and methods, leaving the non-linear residue intractable for suppression. Though, some non-linear processing methods have already be raised, they are complicated and inefficient for suppression, and would bring damage to the speech audio. In this paper, a fusion scheme by combining adaptive filter and neural network is proposed for AEC. The echo could be reduced in a large scale by adaptive filtering, resulting in little residual echo. Though it is much smaller than speech audio, it could also be perceived by human ear and would make communication annoy. The neural network is elaborately designed and trained for suppressing such residual echo. Experiments compared with prevailing methods are conducted, validating the effectiveness and superiority of the proposed combination scheme.

Lu Ma, Xiaomeng Zhang, Pei Zhao et al.

Due to the superior modeling ability of deep neural network (DNN), it is widely used in voice activity detection (VAD). However, the performance may degrade if no sufficient data especially for practical data could be used for training, thus, leading to inferior ability of adaption to environment. Moreover, large model structure could not always be used in practical, especially for low cost devices where restricted hardware is used. This is on the contrary for Gaussian mixture model (GMM) where model parameters can be updated in real-time, but, with low modeling ability. In this paper, deeply integrated scheme combining these two models are proposed to improve adaptability and modeling ability. This is done by directly combining the results of models and feeding it back, together with the result of the DNN model, to update the GMM model. Besides, a control scheme is elaborately designed to detect the endpoints of speech. The superior performance by employing this scheme is validated through experiments in practical, which give an insight into the advantage of combining supervised learning and unsupervised learning.

Lu Ma, Mahsa Ghafarianzadeh, Dave Coleman et al.

We present an unsupervised framework for simultaneous appearance-based object discovery, detection, tracking and reconstruction using RGBD cameras and a robot manipulator. The system performs dense 3D simultaneous localization and mapping concurrently with unsupervised object discovery. Putative objects that are spatially and visually coherent are manipulated by the robot to gain additional motion-cues. The robot uses appearance alone, followed by structure and motion cues, to jointly discover, verify, learn and improve models of objects. Induced motion segmentation reinforces learned models which are represented implicitly as 2D and 3D level sets to capture both shape and appearance. We compare three different approaches for appearance-based object discovery and find that a novel form of spatio-temporal super-pixels gives the highest quality candidate object models in terms of precision and recall. Live experiments with a Baxter robot demonstrate a holistic pipeline capable of automatic discovery, verification, detection, tracking and reconstruction of unknown objects.