Jiahui Geng, Fengyu Cai, Shaobo Cui, Qing Li, Liangwei Chen, Chenyang Lyu, Haonan Li, Derui Zhu, Alexander Pretschner, Heinz Koeppl, Fakhri Karray. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

The spectral shape holds crucial information for Audio Classification (AC), encompassing the spectrum's envelope, details, and dynamic changes over time. Conventional methods utilize cepstral coefficients for spectral shape description but overlook its variation details. Deep-learning approaches capture some dynamics but demand substantial training or fine-tuning resources. The Learning in the Model Space (LMS) framework precisely captures the dynamic information of temporal data by utilizing model fitting, even when computational resources and data are limited. However, applying LMS to audio faces challenges: 1) The high sampling rate of audio hinders efficient data fitting and capturing of dynamic information. 2) The Dynamic Information of Partial Spectral Shapes (DIPSS) may enhance classification, as only specific spectral shapes are relevant for AC. This paper extends an AC framework called Effective Dynamic Information Capture (EDIC) to tackle the above issues. EDIC constructs Mel-Frequency Cepstral Coefficients (MFCC) sequences within different dimensional intervals as the fitted data, which not only reduces the number of sequence sampling points but can also describe the change of the spectral shape in different parts over time. EDIC enables us to implement a topology-based selection algorithm in the model space, selecting effective DIPSS for the current AC task. The performance on three tasks confirms the effectiveness of EDIC.

With the rapid growth of audio data, there's a pressing need for automatic audio classification. As a type of time-series data, audio exhibits waveform fluctuations in both the time and frequency domains that evolve over time, with similar instances sharing consistent patterns. This study introduces the Audio Scanning Network (ASNet), designed to leverage abundant information for achieving stable and effective audio classification. ASNet captures real-time changes in audio waveforms across both time and frequency domains through reservoir computing, supported by Reservoir Kernel Canonical Correlation Analysis (RKCCA) to explore correlations between time-domain and frequency-domain waveform fluctuations. This innovative approach empowers ASNet to comprehensively capture the changes and inherent correlations within the audio waveform, and without the need for time-consuming iterative training. Instead of converting audio into spectrograms, ASNet directly utilizes audio feature sequences to uncover associations between time and frequency fluctuations. Experiments on environmental sound and music genre classification tasks demonstrate ASNet's comparable performance to state-of-the-art methods.

How to learn discriminative video representation from unlabeled videos is challenging but crucial for video analysis. The latest attempts seek to learn a representation model by predicting the appearance contents in the masked regions. However, simply masking and recovering appearance contents may not be sufficient to model temporal clues as the appearance contents can be easily reconstructed from a single frame. To overcome this limitation, we present Masked Motion Encoding (MME), a new pretraining paradigm that reconstructs both appearance and motion information to explore temporal clues. In MME, we focus on addressing two critical challenges to improve the representation performance: 1) how to well represent the possible long-term motion across multiple frames; and 2) how to obtain fine-grained temporal clues from sparsely sampled videos. Motivated by the fact that human is able to recognize an action by tracking objects' position changes and shape changes, we propose to reconstruct a motion trajectory that represents these two kinds of change in the masked regions. Besides, given the sparse video input, we enforce the model to reconstruct dense motion trajectories in both spatial and temporal dimensions. Pre-trained with our MME paradigm, the model is able to anticipate long-term and fine-grained motion details. Code is available at https://github.com/XinyuSun/MME.

In surveys, it is typically up to the individuals to decide if they want to participate or not, which leads to participation bias: the individuals willing to share their data might not be representative of the entire population. Similarly, there are cases where one does not have direct access to any data of the target population and has to resort to publicly available proxy data sampled from a different distribution. In this paper, we present Differentially Private Propensity Scores for Bias Correction (DiPPS), a method for approximating the true data distribution of interest in both of the above settings. We assume that the data analyst has access to a dataset D' that was sampled from the distribution of interest in a biased way. As individuals may be more willing to share their data when given a privacy guarantee, we further assume that the analyst is allowed locally differentially private access to a set of samples D from the true, unbiased distribution. Each data point from the private, unbiased dataset D is mapped to a probability distribution over clusters (learned from the biased dataset D'), from which a single cluster is sampled via the exponential mechanism and shared with the data analyst. This way, the analyst gathers a distribution over clusters, which they use to compute propensity scores for the points in the biased D', which are in turn used to reweight the points in D' to approximate the true data distribution. It is now possible to compute any function on the resulting reweighted dataset without further access to the private D. In experiments on datasets from various domains, we show that DiPPS successfully brings the distribution of the available dataset closer to the distribution of interest in terms of Wasserstein distance. We further show that this results in improved estimates for different statistics, in many cases even outperforming differential privacy mechanisms that are specifically designed for these statistics.

Knowledge-based dialog systems have attracted increasing research interest in diverse applications. However, for disease diagnosis, the widely used knowledge graph (KG) is hard to represent the symptom-symptom and symptom-disease relations since the edges of traditional KG are unweighted. Most research on disease diagnosis dialog systems highly relies on data-driven methods and statistical features, lacking profound comprehension of symptom-symptom and symptom-disease relations. To tackle this issue, this work presents a weighted heterogeneous graph-based dialog system for disease diagnosis. Specifically, we build a weighted heterogeneous graph based on symptom co-occurrence and the proposed symptom frequency-inverse disease frequency. Then, this work proposes a graph-based deep Q -network (graph-DQN) for dialog management. By combining graph convolutional network (GCN) with DQN to learn the embeddings of diseases and symptoms from both the structural and attribute information in the weighted heterogeneous graph, graph-DQN could capture the symptom-disease relations and symptom-symptom relations better. Experimental results show that the proposed dialog system rivals the state-of-the-art models. More importantly, the proposed dialog system can complete the task with fewer dialog turns and possess a better distinguishing capability on diseases with similar symptoms.