Rui Xia

Zengzhi Wang, Xuefeng Li, Rui Xia et al.

High-quality, large-scale corpora are the cornerstone of building foundation models. In this work, we introduce MathPile, a diverse and high-quality math-centric corpus comprising about 9.5 billion tokens. Throughout its creation, we adhered to the principle of "less is more", firmly believing in the supremacy of data quality over quantity, even in the pre-training phase. Our meticulous data collection and processing efforts included a complex suite of preprocessing, prefiltering, language identification, cleaning, filtering, and deduplication, ensuring the high quality of our corpus. Furthermore, we performed data contamination detection on downstream benchmark test sets to eliminate duplicates and conducted continual pre-training experiments, booting the performance on common mathematical reasoning benchmarks. We aim for our MathPile to boost language models' mathematical reasoning abilities and open-source its different versions and processing scripts to advance the field.

Tengfei Wang, Zongqian Zhan, Rui Xia et al.

Over the last few decades, image-based building surface reconstruction has garnered substantial research interest and has been applied across various fields, such as heritage preservation, architectural planning, etc. Compared to the traditional photogrammetric and NeRF-based solutions, recently, Gaussian fields-based methods have exhibited significant potential in generating surface meshes due to their time-efficient training and detailed 3D information preservation. However, most gaussian fields-based methods are trained with all image pixels, encompassing building and nonbuilding areas, which results in a significant noise for building meshes and degeneration in time efficiency. This paper proposes a novel framework, Masked Gaussian Fields (MGFs), designed to generate accurate surface reconstruction for building in a time-efficient way. The framework first applies EfficientSAM and COLMAP to generate multi-level masks of building and the corresponding masked point clouds. Subsequently, the masked gaussian fields are trained by integrating two innovative losses: a multi-level perceptual masked loss focused on constructing building regions and a boundary loss aimed at enhancing the details of the boundaries between different masks. Finally, we improve the tetrahedral surface mesh extraction method based on the masked gaussian spheres. Comprehensive experiments on UAV images demonstrate that, compared to the traditional method and several NeRF-based and Gaussian-based SOTA solutions, our approach significantly improves both the accuracy and efficiency of building surface reconstruction. Notably, as a byproduct, there is an additional gain in the novel view synthesis of building.

Rui Xia, Dan Jiang, Quan Zhang et al.

Temporal Action Localization (TAL) has garnered significant attention in information retrieval. Existing supervised or weakly supervised methods heavily rely on labeled temporal boundaries and action categories, which are labor-intensive and time-consuming. Consequently, unsupervised temporal action localization (UTAL) has gained popularity. However, current methods face two main challenges: 1) Classification pre-trained features overly focus on highly discriminative regions; 2) Solely relying on visual modality information makes it difficult to determine contextual boundaries. To address these issues, we propose a CLIP-assisted cross-view audiovisual enhanced UTAL method. Specifically, we introduce visual language pre-training (VLP) and classification pre-training-based collaborative enhancement to avoid excessive focus on highly discriminative regions; we also incorporate audio perception to provide richer contextual boundary information. Finally, we introduce a self-supervised cross-view learning paradigm to achieve multi-view perceptual enhancement without additional annotations. Extensive experiments on two public datasets demonstrate our model's superiority over several state-of-the-art competitors.

Zhen Wu, Chengcan Ying, Fei Zhao et al.

Aspect-oriented Fine-grained Opinion Extraction (AFOE) aims at extracting aspect terms and opinion terms from review in the form of opinion pairs or additionally extracting sentiment polarity of aspect term to form opinion triplet. Because of containing several opinion factors, the complete AFOE task is usually divided into multiple subtasks and achieved in the pipeline. However, pipeline approaches easily suffer from error propagation and inconvenience in real-world scenarios. To this end, we propose a novel tagging scheme, Grid Tagging Scheme (GTS), to address the AFOE task in an end-to-end fashion only with one unified grid tagging task. Additionally, we design an effective inference strategy on GTS to exploit mutual indication between different opinion factors for more accurate extractions. To validate the feasibility and compatibility of GTS, we implement three different GTS models respectively based on CNN, BiLSTM, and BERT, and conduct experiments on the aspect-oriented opinion pair extraction and opinion triplet extraction datasets. Extensive experimental results indicate that GTS models outperform strong baselines significantly and achieve state-of-the-art performance.

Dongyang Dai, Li Chen, Yuping Wang et al.

With the popularity of deep neural network, speech synthesis task has achieved significant improvements based on the end-to-end encoder-decoder framework in the recent days. More and more applications relying on speech synthesis technology have been widely used in our daily life. Robust speech synthesis model depends on high quality and customized data which needs lots of collecting efforts. It is worth investigating how to take advantage of low-quality and low resource voice data which can be easily obtained from the Internet for usage of synthesizing personalized voice. In this paper, the proposed end-to-end speech synthesis model uses both speaker embedding and noise representation as conditional inputs to model speaker and noise information respectively. Firstly, the speech synthesis model is pre-trained with both multi-speaker clean data and noisy augmented data; then the pre-trained model is adapted on noisy low-resource new speaker data; finally, by setting the clean speech condition, the model can synthesize the new speaker's clean voice. Experimental results show that the speech generated by the proposed approach has better subjective evaluation results than the method directly fine-tuning pre-trained multi-speaker speech synthesis model with denoised new speaker data.

Rui Xia, Mengran Zhang, Zixiang Ding

The emotion cause extraction (ECE) task aims at discovering the potential causes behind a certain emotion expression in a document. Techniques including rule-based methods, traditional machine learning methods and deep neural networks have been proposed to solve this task. However, most of the previous work considered ECE as a set of independent clause classification problems and ignored the relations between multiple clauses in a document. In this work, we propose a joint emotion cause extraction framework, named RNN-Transformer Hierarchical Network (RTHN), to encode and classify multiple clauses synchronously. RTHN is composed of a lower word-level encoder based on RNNs to encode multiple words in each clause, and an upper clause-level encoder based on Transformer to learn the correlation between multiple clauses in a document. We furthermore propose ways to encode the relative position and global predication information into Transformer that can capture the causality between clauses and make RTHN more efficient. We finally achieve the best performance among 12 compared systems and improve the F1 score of the state-of-the-art from 72.69\% to 76.77\%.

Zixiang Ding, Huihui He, Mengran Zhang et al.

Emotion cause identification aims at identifying the potential causes that lead to a certain emotion expression in text. Several techniques including rule based methods and traditional machine learning methods have been proposed to address this problem based on manually designed rules and features. More recently, some deep learning methods have also been applied to this task, with the attempt to automatically capture the causal relationship of emotion and its causes embodied in the text. In this work, we find that in addition to the content of the text, there are another two kinds of information, namely relative position and global labels, that are also very important for emotion cause identification. To integrate such information, we propose a model based on the neural network architecture to encode the three elements ($i.e.$, text content, relative position and global label), in an unified and end-to-end fashion. We introduce a relative position augmented embedding learning algorithm, and transform the task from an independent prediction problem to a reordered prediction problem, where the dynamic global label information is incorporated. Experimental results on a benchmark emotion cause dataset show that our model achieves new state-of-the-art performance and performs significantly better than a number of competitive baselines. Further analysis shows the effectiveness of the relative position augmented embedding learning algorithm and the reordered prediction mechanism with dynamic global labels.

Shiliang Zheng, Rui Xia

Deep learning techniques have achieved success in aspect-based sentiment analysis in recent years. However, there are two important issues that still remain to be further studied, i.e., 1) how to efficiently represent the target especially when the target contains multiple words; 2) how to utilize the interaction between target and left/right contexts to capture the most important words in them. In this paper, we propose an approach, called left-center-right separated neural network with rotatory attention (LCR-Rot), to better address the two problems. Our approach has two characteristics: 1) it has three separated LSTMs, i.e., left, center and right LSTMs, corresponding to three parts of a review (left context, target phrase and right context); 2) it has a rotatory attention mechanism which models the relation between target and left/right contexts. The target2context attention is used to capture the most indicative sentiment words in left/right contexts. Subsequently, the context2target attention is used to capture the most important word in the target. This leads to a two-side representation of the target: left-aware target and right-aware target. We compare our approach on three benchmark datasets with ten related methods proposed recently. The results show that our approach significantly outperforms the state-of-the-art techniques.

Huihui He, Rui Xia

Recently the deep learning techniques have achieved success in multi-label classification due to its automatic representation learning ability and the end-to-end learning framework. Existing deep neural networks in multi-label classification can be divided into two kinds: binary relevance neural network (BRNN) and threshold dependent neural network (TDNN). However, the former needs to train a set of isolate binary networks which ignore dependencies between labels and have heavy computational load, while the latter needs an additional threshold function mechanism to transform the multi-class probabilities to multi-label outputs. In this paper, we propose a joint binary neural network (JBNN), to address these shortcomings. In JBNN, the representation of the text is fed to a set of logistic functions instead of a softmax function, and the multiple binary classifications are carried out synchronously in one neural network framework. Moreover, the relations between labels are captured via training on a joint binary cross entropy (JBCE) loss. To better meet multi-label emotion classification, we further proposed to incorporate the prior label relations into the JBCE loss. The experimental results on the benchmark dataset show that our model performs significantly better than the state-of-the-art multi-label emotion classification methods, in both classification performance and computational efficiency.

Qingqing Zhou, Rui Xia, Chengzhi Zhang

With the development of e-commerce, many products are now being sold worldwide, and manufacturers are eager to obtain a better understanding of customer behavior in various regions. To achieve this goal, most previous efforts have focused mainly on questionnaires, which are time-consuming and costly. The tremendous volume of product reviews on e-commerce websites has seen a new trend emerge, whereby manufacturers attempt to understand user preferences by analyzing online reviews. Following this trend, this paper addresses the problem of studying customer behavior by exploiting recently developed opinion mining techniques. This work is novel for three reasons. First, questionnaire-based investigation is automatically enabled by employing algorithms for template-based question generation and opinion mining-based answer extraction. Using this system, manufacturers are able to obtain reports of customer behavior featuring a much larger sample size, more direct information, a higher degree of automation, and a lower cost. Second, international customer behavior study is made easier by integrating tools for multilingual opinion mining. Third, this is the first time an automatic questionnaire investigation has been conducted to compare customer behavior in China and America, where product reviews are written and read in Chinese and English, respectively. Our study on digital cameras, smartphones, and tablet computers yields three findings. First, Chinese customers follow the Doctrine of the Mean, and often use euphemistic expressions, while American customers express their opinions more directly. Second, Chinese customers care more about general feelings, while American customers pay more attention to product details. Third, Chinese customers focus on external features, while American customers care more about the internal features of products.