Zixing Zhang

OneRec Team, Biao Yang, Boyang Ding et al.

Generative recommendation models in the OneRec family have been widely deployed in many real-world services, such as short-video, live-streaming, advertising, and e-commerce. However, these generative models can only benefit from the scaling advantage, while their reasoning ability is hard to activate, since we cannot construct meaningful Chain-of-Thought (CoT) sequences consisting of itemic tokens only. Inspired by the success of the reasoning-style ``think before answer'' paradigm in the LLM field, we conduct preliminary studies (i.e., OneRec-Think, OpenOneRec) to explore reasoning capability in generative recommendation. Nevertheless, we notice an unexpected phenomenon: the thinking mode does not show advantages over the non-thinking mode. Drawing insights from recent findings on CoT robustness in multi-modal language models, we argue that effective reasoning in recommendation rests on two factors: perception, the ability to ground itemic tokens in their underlying language semantics, and cognition, the ability to reorganize a user's behavior sequence into coherent latent interest points. We therefore propose OneReason, which includes: (1) strong itemic token perception in pre-training, (2) a three-level cognition-enhanced CoT format for recommendation tasks in SFT, and (3) a specialize-then-unify training recipe in RL to enhance the thinking ability.

Zixing Zhang, Liyizhe Peng, Tao Pang et al.

After the inception of emotion recognition or affective computing, it has increasingly become an active research topic due to its broad applications. Over the past couple of decades, emotion recognition models have gradually migrated from statistically shallow models to neural network-based deep models, which can significantly boost the performance of emotion recognition models and consistently achieve the best results on different benchmarks. Therefore, in recent years, deep models have always been considered the first option for emotion recognition. However, the debut of large language models (LLMs), such as ChatGPT, has remarkably astonished the world due to their emerged capabilities of zero/few-shot learning, in-context learning, chain-of-thought, and others that are never shown in previous deep models. In the present paper, we comprehensively investigate how the LLMs perform in emotion recognition in terms of diverse aspects, including in-context learning, few-short learning, accuracy, generalisation, and explanation. Moreover, we offer some insights and pose other potential challenges, hoping to ignite broader discussions about enhancing emotion recognition in the new era of advanced and generalised large models.

Meishu Song, Zijiang Yang, Andreas Triantafyllopoulos et al.

We propose a novel Dynamic Restrained Uncertainty Weighting Loss to experimentally handle the problem of balancing the contributions of multiple tasks on the ICML ExVo 2022 Challenge. The multitask aims to recognize expressed emotions and demographic traits from vocal bursts jointly. Our strategy combines the advantages of Uncertainty Weight and Dynamic Weight Average, by extending weights with a restraint term to make the learning process more explainable. We use a lightweight multi-exit CNN architecture to implement our proposed loss approach. The experimental H-Mean score (0.394) shows a substantial improvement over the baseline H-Mean score (0.335).

Liyizhe Peng, Zixing Zhang, Tao Pang et al.

The advent of large language models (LLMs) has gained tremendous attention over the past year. Previous studies have shown the astonishing performance of LLMs not only in other tasks but also in emotion recognition in terms of accuracy, universality, explanation, robustness, few/zero-shot learning, and others. Leveraging the capability of LLMs inevitably becomes an essential solution for emotion recognition. To this end, we further comprehensively investigate how LLMs perform in linguistic emotion recognition if we concentrate on this specific task. Specifically, we exemplify a publicly available and widely used LLM -- Chat General Language Model, and customise it for our target by using two different modal adaptation techniques, i.e., deep prompt tuning and low-rank adaptation. The experimental results obtained on six widely used datasets present that the adapted LLM can easily outperform other state-of-the-art but specialised deep models. This indicates the strong transferability and feasibility of LLMs in the field of emotion recognition.

Tao Zhou, Lingyu Shu, Zixing Zhang et al.

Deep learning has shown great promise in physiological signal analysis, yet its progress is hindered by heterogeneous data formats, inconsistent preprocessing strategies, fragmented model pipelines, and non-reproducible experimental setups. To address these limitations, we present Tyee, a unified, modular, and fully-integrated configurable toolkit designed for intelligent physiological healthcare. Tyee introduces three key innovations: (1) a unified data interface and configurable preprocessing pipeline for 12 kinds of signal modalities; (2) a modular and extensible architecture enabling flexible integration and rapid prototyping across tasks; and (3) end-to-end workflow configuration, promoting reproducible and scalable experimentation. Tyee demonstrates consistent practical effectiveness and generalizability, outperforming or matching baselines across all evaluated tasks (with state-of-the-art results on 12 of 13 datasets). The Tyee toolkit is released at https://github.com/SmileHnu/Tyee and actively maintained.

Boyang Ding, Chenglong Chu, Dunju Zang et al.

Autoregressive large language models (LLMs) scale well by expressing diverse tasks as sequences of discrete natural-language tokens and training with next-token prediction, which unifies comprehension and generation under self-supervision. Extending this paradigm to multimodal data requires a shared, discrete representation across modalities. However, most vision-language models (VLMs) still rely on a hybrid interface: discrete text tokens paired with continuous Vision Transformer (ViT) features. Because supervision is largely text-driven, these models are often biased toward understanding and cannot fully leverage large-scale self-supervised learning on non-text data. Recent work has explored discrete visual tokenization to enable fully autoregressive multimodal modeling, showing promising progress toward unified understanding and generation. Yet existing discrete vision tokens frequently lose information due to limited code capacity, resulting in noticeably weaker understanding than continuous-feature VLMs. We present Kelix, a fully discrete autoregressive unified model that closes the understanding gap between discrete and continuous visual representations.

Yuanchao Li, Zixing Zhang, Jing Han et al.

The lack of labeled data is a common challenge in speech classification tasks, particularly those requiring extensive subjective assessment, such as cognitive state classification. In this work, we propose a Semi-Supervised Learning (SSL) framework, introducing a novel multi-view pseudo-labeling method that leverages both acoustic and linguistic characteristics to select the most confident data for training the classification model. Acoustically, unlabeled data are compared to labeled data using the Frechet audio distance, calculated from embeddings generated by multiple audio encoders. Linguistically, large language models are prompted to revise automatic speech recognition transcriptions and predict labels based on our proposed task-specific knowledge. High-confidence data are identified when pseudo-labels from both sources align, while mismatches are treated as low-confidence data. A bimodal classifier is then trained to iteratively label the low-confidence data until a predefined criterion is met. We evaluate our SSL framework on emotion recognition and dementia detection tasks. Experimental results demonstrate that our method achieves competitive performance compared to fully supervised learning using only 30% of the labeled data and significantly outperforms two selected baselines.

Huan Zhao, Yiming Liu, Jina Yao et al.

Recent breakthroughs in single-cell technology have ushered in unparalleled opportunities to decode the molecular intricacy of intricate biological systems, especially those linked to diseases unique to humans. However, these progressions have also ushered in novel obstacles-specifically, the efficient annotation of extensive, long-tailed single-cell data pertaining to disease conditions. To effectively surmount this challenge, we introduce Celler, a state-of-the-art generative pre-training model crafted specifically for the annotation of single-cell data. Celler incorporates two groundbreaking elements: First, we introduced the Gaussian Inflation (GInf) Loss function. By dynamically adjusting sample weights, GInf Loss significantly enhances the model's ability to learn from rare categories while reducing the risk of overfitting for common categories. Secondly, we introduce an innovative Hard Data Mining (HDM) strategy into the training process, specifically targeting the challenging-to-learn minority data samples, which significantly improved the model's predictive accuracy. Additionally, to further advance research in this field, we have constructed a large-scale single-cell dataset: Celler-75, which encompasses 40 million cells distributed across 80 human tissues and 75 specific diseases. This dataset provides critical support for comprehensively exploring the potential of single-cell technology in disease research. Our code is available at https://github.com/AI4science-ym/HiCeller.

Zhongren Dong, Haotian Guo, Weixiang Xu et al.

Neuropsychiatric disorders, such as Alzheimer's disease (AD), depression, and autism spectrum disorder (ASD), are characterized by linguistic and acoustic abnormalities, offering potential biomarkers for early detection. Despite the promise of multi-modal approaches, challenges like multi-lingual generalization and the absence of a unified evaluation framework persist. To address these gaps, we propose FEND (Foundation model-based Evaluation of Neuropsychiatric Disorders), a comprehensive multi-modal framework integrating speech and text modalities for detecting AD, depression, and ASD across the lifespan. Leveraging 13 multi-lingual datasets spanning English, Chinese, Greek, French, and Dutch, we systematically evaluate multi-modal fusion performance. Our results show that multi-modal fusion excels in AD and depression detection but underperforms in ASD due to dataset heterogeneity. We also identify modality imbalance as a prevalent issue, where multi-modal fusion fails to surpass the best mono-modal models. Cross-corpus experiments reveal robust performance in task- and language-consistent scenarios but noticeable degradation in multi-lingual and task-heterogeneous settings. By providing extensive benchmarks and a detailed analysis of performance-influencing factors, FEND advances the field of automated, lifespan-inclusive, and multi-lingual neuropsychiatric disorder assessment. We encourage researchers to adopt the FEND framework for fair comparisons and reproducible research.

Bin Wang, Yang Xu, Huan Zhao et al.

Speech-driven 3D talking head generation aims to produce lifelike facial animations precisely synchronized with speech. While considerable progress has been made in achieving high lip-synchronization accuracy, existing methods largely overlook the intricate nuances of individual speaking styles, which limits personalization and realism. In this work, we present a novel framework for personalized 3D talking head animation, namely "PTalker". This framework preserves speaking style through style disentanglement from audio and facial motion sequences and enhances lip-synchronization accuracy through a three-level alignment mechanism between audio and mesh modalities. Specifically, to effectively disentangle style and content, we design disentanglement constraints that encode driven audio and motion sequences into distinct style and content spaces to enhance speaking style representation. To improve lip-synchronization accuracy, we adopt a modality alignment mechanism incorporating three aspects: spatial alignment using Graph Attention Networks to capture vertex connectivity in the 3D mesh structure, temporal alignment using cross-attention to capture and synchronize temporal dependencies, and feature alignment by top-k bidirectional contrastive losses and KL divergence constraints to ensure consistency between speech and mesh modalities. Extensive qualitative and quantitative experiments on public datasets demonstrate that PTalker effectively generates realistic, stylized 3D talking heads that accurately match identity-specific speaking styles, outperforming state-of-the-art methods. The source code and supplementary videos are available at: PTalker.

Haotian Guo, Jing Han, Yongfeng Tu et al.

Despite extensive research on textual and visual disambiguation, disambiguation through speech (DTS) remains underexplored. This is largely due to the lack of high-quality datasets that pair spoken sentences with richly ambiguous text. To address this gap, we present DEBATE, a unique public Chinese speech-text dataset designed to study how speech cues and patterns-pronunciation, pause, stress and intonation-can help resolve textual ambiguity and reveal a speaker's true intent. DEBATE contains 1,001 carefully selected ambiguous utterances, each recorded by 10 native speakers, capturing diverse linguistic ambiguities and their disambiguation through speech. We detail the data collection pipeline and provide rigorous quality analysis. Additionally, we benchmark three state-of-the-art large speech and audio-language models, illustrating clear and huge performance gaps between machine and human understanding of spoken intent. DEBATE represents the first effort of its kind and offers a foundation for building similar DTS datasets across languages and cultures. The dataset and associated code are available at: https://github.com/SmileHnu/DEBATE.

Zhiqiang Gao, Shihao Gao, Zixing Zhang et al.

Understanding sentiment in multimodal conversations is a complex yet crucial challenge toward building emotionally intelligent AI systems. The Multimodal Conversational Aspect-based Sentiment Analysis (MCABSA) Challenge invited participants to tackle two demanding subtasks: (1) extracting a comprehensive sentiment sextuple, including holder, target, aspect, opinion, sentiment, and rationale from multi-speaker dialogues, and (2) detecting sentiment flipping, which detects dynamic sentiment shifts and their underlying triggers. For Subtask-I, in the present paper, we designed a structured prompting pipeline that guided large language models (LLMs) to sequentially extract sentiment components with refined contextual understanding. For Subtask-II, we further leveraged the complementary strengths of three LLMs through ensembling to robustly identify sentiment transitions and their triggers. Our system achieved a 47.38% average score on Subtask-I and a 74.12% exact match F1 on Subtask-II, showing the effectiveness of step-wise refinement and ensemble strategies in rich, multimodal sentiment analysis tasks.

Tianyi Gong, Zijian Cao, Zixing Zhang et al.

Video streaming is a fundamental Internet service, while the quality still cannot be guaranteed especially in poor network conditions such as bandwidth-constrained and remote areas. Existing works mainly work towards two directions: traditional pixel-codec streaming nearly approaches its limit and is hard to step further in compression; the emerging neural-enhanced or generative streaming usually fall short in latency and visual fidelity, hindering their practical deployment. Inspired by the recent success of vision foundation model (VFM), we strive to harness the powerful video understanding and processing capacities of VFM to achieve generalization, high fidelity and loss resilience for real-time video streaming with even higher compression rate. We present the first revolutionized paradigm that enables VFM-based end-to-end generative video streaming towards this goal. Specifically, Morphe employs joint training of visual tokenizers and variable-resolution spatiotemporal optimization under simulated network constraints. Additionally, a robust streaming system is constructed that leverages intelligent packet dropping to resist real-world network perturbations. Extensive evaluation demonstrates that Morphe achieves comparable visual quality while saving 62.5\% bandwidth compared to H.265, and accomplishes real-time, loss-resilient video delivery in challenging network environments, representing a milestone in VFM-enabled multimedia streaming solutions.

Chenglong Chu, Guorui Zhou, Guowang Zhang et al.

Long-context ability, has become one of the most important iteration direction of next-generation Large Language Models, particularly in semantic understanding/reasoning, code agentic intelligence and recommendation system. However, the standard softmax attention exhibits quadratic time complexity with respect to sequence length. As the sequence length increases, this incurs substantial overhead in long-context settings, leading the training and inference costs of extremely long sequences deteriorate rapidly. Existing solutions mitigate this issue through two technique routings: i) Reducing the KV cache per layer, such as from the head-level compression GQA, and the embedding dimension-level compression MLA, but the KV cache remains linearly dependent on the sequence length at a 1:1 ratio. ii) Interleaving with KV Cache friendly architecture, such as local attention SWA, linear kernel GDN, but often involve trade-offs among KV Cache and long-context modeling effectiveness. Besides the two technique routings, we argue that there exists an intermediate path not well explored: {Maintaining a linear relationship between the KV cache and sequence length, but performing semantic-level compression through a specific ratio $k$}. This $O(n/k)$ path does not pursue a ``minimum KV cache'', but rather trades acceptable memory costs for complete, referential, and interpretable retention of long distant dependency. Motivated by this, we propose Kwai Summary Attention (KSA), a novel attention mechanism that reduces sequence modeling cost by compressing historical contexts into learnable summary tokens.

Kwai Keye Team, Biao Yang, Bin Wen et al.

While Multimodal Large Language Models (MLLMs) demonstrate remarkable capabilities on static images, they often fall short in comprehending dynamic, information-dense short-form videos, a dominant medium in today's digital landscape. To bridge this gap, we introduce \textbf{Kwai Keye-VL}, an 8-billion-parameter multimodal foundation model engineered for leading-edge performance in short-video understanding while maintaining robust general-purpose vision-language abilities. The development of Keye-VL rests on two core pillars: a massive, high-quality dataset exceeding 600 billion tokens with a strong emphasis on video, and an innovative training recipe. This recipe features a four-stage pre-training process for solid vision-language alignment, followed by a meticulous two-phase post-training process. The first post-training stage enhances foundational capabilities like instruction following, while the second phase focuses on stimulating advanced reasoning. In this second phase, a key innovation is our five-mode ``cold-start'' data mixture, which includes ``thinking'', ``non-thinking'', ``auto-think'', ``think with image'', and high-quality video data. This mixture teaches the model to decide when and how to reason. Subsequent reinforcement learning (RL) and alignment steps further enhance these reasoning capabilities and correct abnormal model behaviors, such as repetitive outputs. To validate our approach, we conduct extensive evaluations, showing that Keye-VL achieves state-of-the-art results on public video benchmarks and remains highly competitive on general image-based tasks (Figure 1). Furthermore, we develop and release the \textbf{KC-MMBench}, a new benchmark tailored for real-world short-video scenarios, where Keye-VL shows a significant advantage.

Zhongren Dong, Zixing Zhang, Weixiang Xu et al.

Automatically detecting Alzheimer's Disease (AD) from spontaneous speech plays an important role in its early diagnosis. Recent approaches highly rely on the Transformer architectures due to its efficiency in modelling long-range context dependencies. However, the quadratic increase in computational complexity associated with self-attention and the length of audio poses a challenge when deploying such models on edge devices. In this context, we construct a novel framework, namely Hierarchical Attention-Free Transformer (HAFFormer), to better deal with long speech for AD detection. Specifically, we employ an attention-free module of Multi-Scale Depthwise Convolution to replace the self-attention and thus avoid the expensive computation, and a GELU-based Gated Linear Unit to replace the feedforward layer, aiming to automatically filter out the redundant information. Moreover, we design a hierarchical structure to force it to learn a variety of information grains, from the frame level to the dialogue level. By conducting extensive experiments on the ADReSS-M dataset, the introduced HAFFormer can achieve competitive results (82.6% accuracy) with other recent work, but with significant computational complexity and model size reduction compared to the standard Transformer. This shows the efficiency of HAFFormer in dealing with long audio for AD detection.

Biao Yang, Bin Wen, Boyang Ding et al.

In recent years, the development of Large Language Models (LLMs) has significantly advanced, extending their capabilities to multimodal tasks through Multimodal Large Language Models (MLLMs). However, video understanding remains a challenging area due to the dynamic and information-dense nature of videos. Existing models struggle with the trade-off between spatial resolution and temporal coverage when processing video content. We present Keye-VL-1.5, which addresses fundamental challenges in video comprehension through three key innovations. First, we introduce a novel Slow-Fast video encoding strategy that dynamically allocates computational resources based on inter-frame similarity, processing key frames with significant visual changes at higher resolution (Slow pathway) while handling relatively static frames with increased temporal coverage at lower resolution (Fast pathway). Second, we implement a progressive four-stage pre-training methodology that systematically extends the model's context length from 8K to 128K tokens, enabling processing of longer videos and more complex visual content. Third, we develop a comprehensive post-training pipeline focusing on reasoning enhancement and human preference alignment, incorporating a 5-step chain-of-thought data construction process, iterative GSPO-based reinforcement learning with progressive prompt hinting for difficult cases, and alignment training. Through extensive evaluation on public benchmarks and rigorous internal human assessment, Keye-VL-1.5 demonstrates significant improvements over existing models, particularly excelling in video understanding tasks while maintaining competitive performance on general multimodal benchmarks.

Xupeng Zha, Huan Zhao, Zixing Zhang

Conversational Emotion Recognition (CER) aims to predict the emotion expressed by an utterance (referred to as an ``event'') during a conversation. Existing graph-based methods mainly focus on event interactions to comprehend the conversational context, while overlooking the direct influence of the speaker's emotional state on the events. In addition, real-time modeling of the conversation is crucial for real-world applications but is rarely considered. Toward this end, we propose a novel graph-based approach, namely Event-State Interactions infused Heterogeneous Graph Neural Network (ESIHGNN), which incorporates the speaker's emotional state and constructs a heterogeneous event-state interaction graph to model the conversation. Specifically, a heterogeneous directed acyclic graph neural network is employed to dynamically update and enhance the representations of events and emotional states at each turn, thereby improving conversational coherence and consistency. Furthermore, to further improve the performance of CER, we enrich the graph's edges with external knowledge. Experimental results on four publicly available CER datasets show the superiority of our approach and the effectiveness of the introduced heterogeneous event-state interaction graph.

Yi Chang, Zhao Ren, Zixing Zhang et al.

Speech contains rich information on the emotions of humans, and Speech Emotion Recognition (SER) has been an important topic in the area of human-computer interaction. The robustness of SER models is crucial, particularly in privacy-sensitive and reliability-demanding domains like private healthcare. Recently, the vulnerability of deep neural networks in the audio domain to adversarial attacks has become a popular area of research. However, prior works on adversarial attacks in the audio domain primarily rely on iterative gradient-based techniques, which are time-consuming and prone to overfitting the specific threat model. Furthermore, the exploration of sparse perturbations, which have the potential for better stealthiness, remains limited in the audio domain. To address these challenges, we propose a generator-based attack method to generate sparse and transferable adversarial examples to deceive SER models in an end-to-end and efficient manner. We evaluate our method on two widely-used SER datasets, Database of Elicited Mood in Speech (DEMoS) and Interactive Emotional dyadic MOtion CAPture (IEMOCAP), and demonstrate its ability to generate successful sparse adversarial examples in an efficient manner. Moreover, our generated adversarial examples exhibit model-agnostic transferability, enabling effective adversarial attacks on advanced victim models.

Zixing Zhang, Zhongren Dong, Weixiang Xu et al.

With the increasing implementation of machine learning models on edge or Internet-of-Things (IoT) devices, deploying advanced models on resource-constrained IoT devices remains challenging. Transformer models, a currently dominant neural architecture, have achieved great success in broad domains but their complexity hinders its deployment on IoT devices with limited computation capability and storage size. Although many model compression approaches have been explored, they often suffer from notorious performance degradation. To address this issue, we introduce a new method, namely Transformer Re-parameterization, to boost the performance of lightweight Transformer models. It consists of two processes: the High-Rank Factorization (HRF) process in the training stage and the deHigh-Rank Factorization (deHRF) process in the inference stage. In the former process, we insert an additional linear layer before the Feed-Forward Network (FFN) of the lightweight Transformer. It is supposed that the inserted HRF layers can enhance the model learning capability. In the later process, the auxiliary HRF layer will be merged together with the following FFN layer into one linear layer and thus recover the original structure of the lightweight model. To examine the effectiveness of the proposed method, we evaluate it on three widely used Transformer variants, i.e., ConvTransformer, Conformer, and SpeechFormer networks, in the application of speech emotion recognition on the IEMOCAP, M3ED and DAIC-WOZ datasets. Experimental results show that our proposed method consistently improves the performance of lightweight Transformers, even making them comparable to large models. The proposed re-parameterization approach enables advanced Transformer models to be deployed on resource-constrained IoT devices.

Xiangheng He, Junjie Chen, Zixing Zhang et al.

Prosody contains rich information beyond the literal meaning of words, which is crucial for the intelligibility of speech. Current models still fall short in phrasing and intonation; they not only miss or misplace breaks when synthesizing long sentences with complex structures but also produce unnatural intonation. We propose ProsodyFM, a prosody-aware text-to-speech synthesis (TTS) model with a flow-matching (FM) backbone that aims to enhance the phrasing and intonation aspects of prosody. ProsodyFM introduces two key components: a Phrase Break Encoder to capture initial phrase break locations, followed by a Duration Predictor for the flexible adjustment of break durations; and a Terminal Intonation Encoder which learns a bank of intonation shape tokens combined with a novel Pitch Processor for more robust modeling of human-perceived intonation change. ProsodyFM is trained with no explicit prosodic labels and yet can uncover a broad spectrum of break durations and intonation patterns. Experimental results demonstrate that ProsodyFM can effectively improve the phrasing and intonation aspects of prosody, thereby enhancing the overall intelligibility compared to four state-of-the-art (SOTA) models. Out-of-distribution experiments show that this prosody improvement can further bring ProsodyFM superior generalizability for unseen complex sentences and speakers. Our case study intuitively illustrates the powerful and fine-grained controllability of ProsodyFM over phrasing and intonation.

Jing Han, Kun Qian, Meishu Song et al.

The COVID-19 outbreak was announced as a global pandemic by the World Health Organisation in March 2020 and has affected a growing number of people in the past few weeks. In this context, advanced artificial intelligence techniques are brought to the fore in responding to fight against and reduce the impact of this global health crisis. In this study, we focus on developing some potential use-cases of intelligent speech analysis for COVID-19 diagnosed patients. In particular, by analysing speech recordings from these patients, we construct audio-only-based models to automatically categorise the health state of patients from four aspects, including the severity of illness, sleep quality, fatigue, and anxiety. For this purpose, two established acoustic feature sets and support vector machines are utilised. Our experiments show that an average accuracy of .69 obtained estimating the severity of illness, which is derived from the number of days in hospitalisation. We hope that this study can foster an extremely fast, low-cost, and convenient way to automatically detect the COVID-19 disease.

Jing Han, Zixing Zhang, Zhao Ren et al.

Despite remarkable advances in emotion recognition, they are severely restrained from either the essentially limited property of the employed single modality, or the synchronous presence of all involved multiple modalities. Motivated by this, we propose a novel crossmodal emotion embedding framework called EmoBed, which aims to leverage the knowledge from other auxiliary modalities to improve the performance of an emotion recognition system at hand. The framework generally includes two main learning components, i. e., joint multimodal training and crossmodal training. Both of them tend to explore the underlying semantic emotion information but with a shared recognition network or with a shared emotion embedding space, respectively. In doing this, the enhanced system trained with this approach can efficiently make use of the complementary information from other modalities. Nevertheless, the presence of these auxiliary modalities is not demanded during inference. To empirically investigate the effectiveness and robustness of the proposed framework, we perform extensive experiments on the two benchmark databases RECOLA and OMG-Emotion for the tasks of dimensional emotion regression and categorical emotion classification, respectively. The obtained results show that the proposed framework significantly outperforms related baselines in monomodal inference, and are also competitive or superior to the recently reported systems, which emphasises the importance of the proposed crossmodal learning for emotion recognition.

Zixing Zhang, Bingwen Wu, Bjoern Schuller

Despite the increasing research interest in end-to-end learning systems for speech emotion recognition, conventional systems either suffer from the overfitting due in part to the limited training data, or do not explicitly consider the different contributions of automatically learnt representations for a specific task. In this contribution, we propose a novel end-to-end framework which is enhanced by learning other auxiliary tasks and an attention mechanism. That is, we jointly train an end-to-end network with several different but related emotion prediction tasks, i.e., arousal, valence, and dominance predictions, to extract more robust representations shared among various tasks than traditional systems with the hope that it is able to relieve the overfitting problem. Meanwhile, an attention layer is implemented on top of the layers for each task, with the aim to capture the contribution distribution of different segment parts for each individual task. To evaluate the effectiveness of the proposed system, we conducted a set of experiments on the widely used database IEMOCAP. The empirical results show that the proposed systems significantly outperform corresponding baseline systems.

Zixing Zhang, Jing Han, Kun Qian et al.

One of the frontier issues that severely hamper the development of automatic snore sound classification (ASSC) associates to the lack of sufficient supervised training data. To cope with this problem, we propose a novel data augmentation approach based on semi-supervised conditional Generative Adversarial Networks (scGANs), which aims to automatically learn a mapping strategy from a random noise space to original data distribution. The proposed approach has the capability of well synthesizing 'realistic' high-dimensional data, while requiring no additional annotation process. To handle the mode collapse problem of GANs, we further introduce an ensemble strategy to enhance the diversity of the generated data. The systematic experiments conducted on a widely used Munich-Passau snore sound corpus demonstrate that the scGANs-based systems can remarkably outperform other classic data augmentation systems, and are also competitive to other recently reported systems for ASSC.

Zixing Zhang, Jing Han, Eduardo Coutinho et al.

Time-continuous emotion prediction has become an increasingly compelling task in machine learning. Considerable efforts have been made to advance the performance of these systems. Nonetheless, the main focus has been the development of more sophisticated models and the incorporation of different expressive modalities (e. g., speech, face, and physiology). In this paper, motivated by the benefit of difficulty awareness in a human learning procedure, we propose a novel machine learning framework, namely, Dynamic Difficulty Awareness Training (DDAT), which sheds fresh light on the research -- directly exploiting the difficulties in learning to boost the machine learning process. The DDAT framework consists of two stages: information retrieval and information exploitation. In the first stage, we make use of the reconstruction error of input features or the annotation uncertainty to estimate the difficulty of learning specific information. The obtained difficulty level is then used in tandem with original features to update the model input in a second learning stage with the expectation that the model can learn to focus on high difficulty regions of the learning process. We perform extensive experiments on a benchmark database (RECOLA) to evaluate the effectiveness of the proposed framework. The experimental results show that our approach outperforms related baselines as well as other well-established time-continuous emotion prediction systems, which suggests that dynamically integrating the difficulty information for neural networks can help enhance the learning process.

Jing Han, Zixing Zhang, Nicholas Cummins et al.

Over the past few years, adversarial training has become an extremely active research topic and has been successfully applied to various Artificial Intelligence (AI) domains. As a potentially crucial technique for the development of the next generation of emotional AI systems, we herein provide a comprehensive overview of the application of adversarial training to affective computing and sentiment analysis. Various representative adversarial training algorithms are explained and discussed accordingly, aimed at tackling diverse challenges associated with emotional AI systems. Further, we highlight a range of potential future research directions. We expect that this overview will help facilitate the development of adversarial training for affective computing and sentiment analysis in both the academic and industrial communities.

Zixing Zhang, Ding Liu, Jing Han et al.

Acoustic Event Classification (AEC) has become a significant task for machines to perceive the surrounding auditory scene. However, extracting effective representations that capture the underlying characteristics of the acoustic events is still challenging. Previous methods mainly focused on designing the audio features in a `hand-crafted' manner. Interestingly, data-learnt features have been recently reported to show better performance. Up to now, these were only considered on the frame level. In this article, we propose an unsupervised learning framework to learn a vector representation of an audio sequence for AEC. This framework consists of a Recurrent Neural Network (RNN) encoder and an RNN decoder, which respectively transforms the variable-length audio sequence into a fixed-length vector and reconstructs the input sequence on the generated vector. After training the encoder-decoder, we feed the audio sequences to the encoder and then take the learnt vectors as the audio sequence representations. Compared with previous methods, the proposed method can not only deal with the problem of arbitrary-lengths of audio streams, but also learn the salient information of the sequence. Extensive evaluation on a large-size acoustic event database is performed, and the empirical results demonstrate that the learnt audio sequence representation yields a significant performance improvement by a large margin compared with other state-of-the-art hand-crafted sequence features for AEC.

Zixing Zhang, Jürgen Geiger, Jouni Pohjalainen et al.

Eliminating the negative effect of non-stationary environmental noise is a long-standing research topic for automatic speech recognition that stills remains an important challenge. Data-driven supervised approaches, including ones based on deep neural networks, have recently emerged as potential alternatives to traditional unsupervised approaches and with sufficient training, can alleviate the shortcomings of the unsupervised methods in various real-life acoustic environments. In this light, we review recently developed, representative deep learning approaches for tackling non-stationary additive and convolutional degradation of speech with the aim of providing guidelines for those involved in the development of environmentally robust speech recognition systems. We separately discuss single- and multi-channel techniques developed for the front-end and back-end of speech recognition systems, as well as joint front-end and back-end training frameworks.