Jian Yang

Jian Yang, Shaohan Huang, Shuming Ma et al. · microsoft-research

Named entity recognition (NER) suffers from the scarcity of annotated training data, especially for low-resource languages without labeled data. Cross-lingual NER has been proposed to alleviate this issue by transferring knowledge from high-resource languages to low-resource languages via aligned cross-lingual representations or machine translation results. However, the performance of cross-lingual NER methods is severely affected by the unsatisfactory quality of translation or label projection. To address these problems, we propose a Cross-lingual Entity Projection framework (CROP) to enable zero-shot cross-lingual NER with the help of a multilingual labeled sequence translation model. Specifically, the target sequence is first translated into the source language and then tagged by a source NER model. We further adopt a labeled sequence translation model to project the tagged sequence back to the target language and label the target raw sentence. Ultimately, the whole pipeline is integrated into an end-to-end model by the way of self-training. Experimental results on two benchmarks demonstrate that our method substantially outperforms the previous strong baseline by a large margin of +3~7 F1 scores and achieves state-of-the-art performance.

Jian Yang, Shuming Ma, Li Dong et al. · microsoft-research

Pre-trained models have achieved remarkable success in natural language processing (NLP). However, existing pre-training methods underutilize the benefits of language understanding for generation. Inspired by the idea of Generative Adversarial Networks (GANs), we propose a GAN-style model for encoder-decoder pre-training by introducing an auxiliary discriminator, unifying the ability of language understanding and generation in a single model. Our model, named as GanLM, is trained with two pre-training objectives: replaced token detection and replaced token denoising. Specifically, given masked source sentences, the generator outputs the target distribution and the discriminator predicts whether the target sampled tokens from distribution are incorrect. The target sentence is replaced with misclassified tokens to construct noisy previous context, which is used to generate the gold sentence. In general, both tasks improve the ability of language understanding and generation by selectively using the denoising data. Extensive experiments in language generation benchmarks show that GanLM with the powerful language understanding capability outperforms various strong pre-trained language models (PLMs) and achieves state-of-the-art performance.

Hongcheng Guo, Jiaheng Liu, Haoyang Huang et al. · microsoft-research

Multimodal Machine Translation (MMT) focuses on enhancing text-only translation with visual features, which has attracted considerable attention from both natural language processing and computer vision communities. Recent advances still struggle to train a separate model for each language pair, which is costly and unaffordable when the number of languages increases in the real world. In other words, the multilingual multimodal machine translation (Multilingual MMT) task has not been investigated, which aims to handle the aforementioned issues by providing a shared semantic space for multiple languages. Besides, the image modality has no language boundaries, which is superior to bridging the semantic gap between languages. To this end, we first propose the Multilingual MMT task by establishing two new Multilingual MMT benchmark datasets covering seven languages. Then, an effective baseline LVP-M3 using visual prompts is proposed to support translations between different languages, which includes three stages (token encoding, language-aware visual prompt generation, and language translation). Extensive experimental results on our constructed benchmark datasets demonstrate the effectiveness of LVP-M3 method for Multilingual MMT.

Jian Yang, Yuwei Yin, Liqun Yang et al. · microsoft-research

Transformer structure, stacked by a sequence of encoder and decoder network layers, achieves significant development in neural machine translation. However, vanilla Transformer mainly exploits the top-layer representation, assuming the lower layers provide trivial or redundant information and thus ignoring the bottom-layer feature that is potentially valuable. In this work, we propose the Group-Transformer model (GTrans) that flexibly divides multi-layer representations of both encoder and decoder into different groups and then fuses these group features to generate target words. To corroborate the effectiveness of the proposed method, extensive experiments and analytic experiments are conducted on three bilingual translation benchmarks and two multilingual translation tasks, including the IWLST-14, IWLST-17, LDC, WMT-14 and OPUS-100 benchmark. Experimental and analytical results demonstrate that our model outperforms its Transformer counterparts by a consistent gain. Furthermore, it can be successfully scaled up to 60 encoder layers and 36 decoder layers.

Ge Zhang, Jia Wu, Jian Yang et al. · allen-ai

Graph-structured data consisting of objects (i.e., nodes) and relationships among objects (i.e., edges) are ubiquitous. Graph-level learning is a matter of studying a collection of graphs instead of a single graph. Traditional graph-level learning methods used to be the mainstream. However, with the increasing scale and complexity of graphs, Graph-level Neural Networks (GLNNs, deep learning-based graph-level learning methods) have been attractive due to their superiority in modeling high-dimensional data. Thus, a survey on GLNNs is necessary. To frame this survey, we propose a systematic taxonomy covering GLNNs upon deep neural networks, graph neural networks, and graph pooling. The representative and state-of-the-art models in each category are focused on this survey. We also investigate the reproducibility, benchmarks, and new graph datasets of GLNNs. Finally, we conclude future directions to further push forward GLNNs. The repository of this survey is available at https://github.com/GeZhangMQ/Awesome-Graph-level-Neural-Networks.

Jian Yang, Yuwei Yin, Shuming Ma et al. · microsoft-research

Most translation tasks among languages belong to the zero-resource translation problem where parallel corpora are unavailable. Multilingual neural machine translation (MNMT) enables one-pass translation using shared semantic space for all languages compared to the two-pass pivot translation but often underperforms the pivot-based method. In this paper, we propose a novel method, named as Unified Multilingual Multiple teacher-student Model for NMT (UM4). Our method unifies source-teacher, target-teacher, and pivot-teacher models to guide the student model for the zero-resource translation. The source teacher and target teacher force the student to learn the direct source to target translation by the distilled knowledge on both source and target sides. The monolingual corpus is further leveraged by the pivot-teacher model to enhance the student model. Experimental results demonstrate that our model of 72 directions significantly outperforms previous methods on the WMT benchmark.

Jian Yang, Yuwei Yin, Shuming Ma et al. · microsoft-research

Multilingual neural machine translation (MNMT) trained in multiple language pairs has attracted considerable attention due to fewer model parameters and lower training costs by sharing knowledge among multiple languages. Nonetheless, multilingual training is plagued by language interference degeneration in shared parameters because of the negative interference among different translation directions, especially on high-resource languages. In this paper, we propose the multilingual translation model with the high-resource language-specific training (HLT-MT) to alleviate the negative interference, which adopts the two-stage training with the language-specific selection mechanism. Specifically, we first train the multilingual model only with the high-resource pairs and select the language-specific modules at the top of the decoder to enhance the translation quality of high-resource directions. Next, the model is further trained on all available corpora to transfer knowledge from high-resource languages (HRLs) to low-resource languages (LRLs). Experimental results show that HLT-MT outperforms various strong baselines on WMT-10 and OPUS-100 benchmarks. Furthermore, the analytic experiments validate the effectiveness of our method in mitigating the negative interference in multilingual training.

Jian Yang, Yuwei Yin, Shuming Ma et al. · bytedance, microsoft-research

Context-aware neural machine translation aims to use the document-level context to improve translation quality. However, not all words in the context are helpful. The irrelevant or trivial words may bring some noise and distract the model from learning the relationship between the current sentence and the auxiliary context. To mitigate this problem, we propose a novel end-to-end encoder-decoder model with a layer-wise selection mechanism to sift and refine the long document context. To verify the effectiveness of our method, extensive experiments and extra quantitative analysis are conducted on four document-level machine translation benchmarks. The experimental results demonstrate that our model significantly outperforms previous models on all datasets via the soft selection mechanism.

Tongliang Li, Zixiang Wang, Linzheng Chai et al. · tsinghua

Cross-lingual open information extraction aims to extract structured information from raw text across multiple languages. Previous work uses a shared cross-lingual pre-trained model to handle the different languages but underuses the potential of the language-specific representation. In this paper, we propose an effective multi-stage tuning framework called MT4CrossIE, designed for enhancing cross-lingual open information extraction by injecting language-specific knowledge into the shared model. Specifically, the cross-lingual pre-trained model is first tuned in a shared semantic space (e.g., embedding matrix) in the fixed encoder and then other components are optimized in the second stage. After enough training, we freeze the pre-trained model and tune the multiple extra low-rank language-specific modules using mixture-of-LoRAs for model-based cross-lingual transfer. In addition, we leverage two-stage prompting to encourage the large language model (LLM) to annotate the multi-lingual raw data for data-based cross-lingual transfer. The model is trained with multi-lingual objectives on our proposed dataset OpenIE4++ by combing the model-based and data-based transfer techniques. Experimental results on various benchmarks emphasize the importance of aggregating multiple plug-in-and-play language-specific modules and demonstrate the effectiveness of MT4CrossIE in cross-lingual OIE\footnote{\url{https://github.com/CSJianYang/Multilingual-Multimodal-NLP}}.

Lingfeng Yang, Xiang Li, Borui Zhao et al.

Mix-based augmentation has been proven fundamental to the generalization of deep vision models. However, current augmentations only mix samples at the current data batch during training, which ignores the possible knowledge accumulated in the learning history. In this paper, we propose a recursive mixed-sample learning paradigm, termed "RecursiveMix" (RM), by exploring a novel training strategy that leverages the historical input-prediction-label triplets. More specifically, we iteratively resize the input image batch from the previous iteration and paste it into the current batch while their labels are fused proportionally to the area of the operated patches. Further, a consistency loss is introduced to align the identical image semantics across the iterations, which helps the learning of scale-invariant feature representations. Based on ResNet-50, RM largely improves classification accuracy by $\sim$3.2\% on CIFAR100 and $\sim$2.8\% on ImageNet with negligible extra computation/storage costs. In the downstream object detection task, the RM pretrained model outperforms the baseline by 2.1 AP points and surpasses CutMix by 1.4 AP points under the ATSS detector on COCO. In semantic segmentation, RM also surpasses the baseline and CutMix by 1.9 and 1.1 mIoU points under UperNet on ADE20K, respectively. Codes and pretrained models are available at \url{https://github.com/megvii-research/RecursiveMix}.

Fanzhen Liu, Xiaoxiao Ma, Jia Wu et al.

Graph anomaly detection in this paper aims to distinguish abnormal nodes that behave differently from the benign ones accounting for the majority of graph-structured instances. Receiving increasing attention from both academia and industry, yet existing research on this task still suffers from two critical issues when learning informative anomalous behavior from graph data. For one thing, anomalies are usually hard to capture because of their subtle abnormal behavior and the shortage of background knowledge about them, which causes severe anomalous sample scarcity. Meanwhile, the overwhelming majority of objects in real-world graphs are normal, bringing the class imbalance problem as well. To bridge the gaps, this paper devises a novel Data Augmentation-based Graph Anomaly Detection (DAGAD) framework for attributed graphs, equipped with three specially designed modules: 1) an information fusion module employing graph neural network encoders to learn representations, 2) a graph data augmentation module that fertilizes the training set with generated samples, and 3) an imbalance-tailored learning module to discriminate the distributions of the minority (anomalous) and majority (normal) classes. A series of experiments on three datasets prove that DAGAD outperforms ten state-of-the-art baseline detectors concerning various mostly-used metrics, together with an extensive ablation study validating the strength of our proposed modules.

Zhenyu Yang, Ge Zhang, Jia Wu et al.

Graphs have a superior ability to represent relational data, like chemical compounds, proteins, and social networks. Hence, graph-level learning, which takes a set of graphs as input, has been applied to many tasks including comparison, regression, classification, and more. Traditional approaches to learning a set of graphs heavily rely on hand-crafted features, such as substructures. But while these methods benefit from good interpretability, they often suffer from computational bottlenecks as they cannot skirt the graph isomorphism problem. Conversely, deep learning has helped graph-level learning adapt to the growing scale of graphs by extracting features automatically and encoding graphs into low-dimensional representations. As a result, these deep graph learning methods have been responsible for many successes. Yet, there is no comprehensive survey that reviews graph-level learning starting with traditional learning and moving through to the deep learning approaches. This article fills this gap and frames the representative algorithms into a systematic taxonomy covering traditional learning, graph-level deep neural networks, graph-level graph neural networks, and graph pooling. To ensure a thoroughly comprehensive survey, the evolutions, interactions, and communications between methods from four different branches of development are also examined. This is followed by a brief review of the benchmark data sets, evaluation metrics, and common downstream applications. The survey concludes with a broad overview of 12 current and future directions in this booming field.

Xiang Li, Ge Wu, Lingfeng Yang et al.

New knowledge originates from the old. The various types of elements, deposited in the training history, are a large amount of wealth for improving learning deep models. In this survey, we comprehensively review and summarize the topic--``Historical Learning: Learning Models with Learning History'', which learns better neural models with the help of their learning history during its optimization, from three detailed aspects: Historical Type (what), Functional Part (where) and Storage Form (how). To our best knowledge, it is the first survey that systematically studies the methodologies which make use of various historical statistics when training deep neural networks. The discussions with related topics like recurrent/memory networks, ensemble learning, and reinforcement learning are demonstrated. We also expose future challenges of this topic and encourage the community to pay attention to the think of historical learning principles when designing algorithms. The paper list related to historical learning is available at \url{https://github.com/Martinser/Awesome-Historical-Learning.}

King Zhu, Qianbo Zang, Shian Jia et al.

Multimodal Large Language Models (MLLMs) are evaluated on various benchmarks, such as image captioning, visual question answering, and reasoning. However, many of these benchmarks include overly simple or uninformative samples, complicating the effective distinction of different MLLMs' performance. Furthermore, evaluating models across numerous benchmarks incurs a significant computational burden. To address these issues, we propose LIME (Less Is More for MLLM Evaluation), a refined and efficient benchmark curated through a semi-automated pipeline. This pipeline filters out uninformative samples and eliminates answer leakage by focusing on tasks that necessitate image-based understanding. Our experiments indicate that LIME reduces the number of samples by 76% and evaluation time by 77%, while also providing a more effective means of distinguishing the capabilities of different models. Notably, we find that traditional automatic metrics, such as CIDEr, are inadequate for assessing MLLMs' captioning performance; excluding the caption task score yields a more accurate reflection of overall model performance. All code and data are available at https://github.com/kangreen0210/LIME.

Xia Hou, Qifeng Li, Jian Yang et al.

Instruction tuning as an effective technique aligns the outputs of large language models (LLMs) with human preference. But how to generate the seasonal multi-turn dialogues from raw documents for instruction tuning still requires further exploration. In this paper, we present a novel framework named R2S that leverages the CoD-Chain of Dialogue logic to guide large language models (LLMs) in generating knowledge-intensive multi-turn dialogues for instruction tuning. By integrating raw documents from both open-source datasets and domain-specific web-crawled documents into a benchmark K-BENCH, we cover diverse areas such as Wikipedia (English), Science (Chinese), and Artifacts (Chinese). Our approach first decides the logic flow of the current dialogue and then prompts LLMs to produce key phrases for sourcing relevant response content. This methodology enables the creation of the G I NSTRUCT instruction dataset, retaining raw document knowledge within dialoguestyle interactions. Utilizing this dataset, we fine-tune GLLM, a model designed to transform raw documents into structured multi-turn dialogues, thereby injecting comprehensive domain knowledge into the SFT model for enhanced instruction tuning. This work signifies a stride towards refining the adaptability and effectiveness of LLMs in processing and generating more accurate, contextually nuanced responses across various fields.

Kun Wang, Zhiqiang Yan, Huang Tian et al.

Neural Radiance Fields (NeRF) have shown promise in generating realistic novel views from sparse scene images. However, existing NeRF approaches often encounter challenges due to the lack of explicit 3D supervision and imprecise camera poses, resulting in suboptimal outcomes. To tackle these issues, we propose AltNeRF -- a novel framework designed to create resilient NeRF representations using self-supervised monocular depth estimation (SMDE) from monocular videos, without relying on known camera poses. SMDE in AltNeRF masterfully learns depth and pose priors to regulate NeRF training. The depth prior enriches NeRF's capacity for precise scene geometry depiction, while the pose prior provides a robust starting point for subsequent pose refinement. Moreover, we introduce an alternating algorithm that harmoniously melds NeRF outputs into SMDE through a consistence-driven mechanism, thus enhancing the integrity of depth priors. This alternation empowers AltNeRF to progressively refine NeRF representations, yielding the synthesis of realistic novel views. Extensive experiments showcase the compelling capabilities of AltNeRF in generating high-fidelity and robust novel views that closely resemble reality.

Qianru Zhou, Rongzhen Li, Lei Xu et al.

The Internet is the most complex machine humankind has ever built, and how to defense it from intrusions is even more complex. With the ever increasing of new intrusions, intrusion detection task rely on Artificial Intelligence more and more. Interpretability and transparency of the machine learning model is the foundation of trust in AI-driven intrusion detection results. Current interpretation Artificial Intelligence technologies in intrusion detection are heuristic, which is neither accurate nor sufficient. This paper proposed a rigorous interpretable Artificial Intelligence driven intrusion detection approach, based on artificial immune system. Details of rigorous interpretation calculation process for a decision tree model is presented. Prime implicant explanation for benign traffic flow are given in detail as rule for negative selection of the cyber immune system. Experiments are carried out in real-life traffic.

Hongcheng Guo, Jian Yang, Jiaheng Liu et al.

With the rapid development of IT operations, it has become increasingly crucial to efficiently manage and analyze large volumes of data for practical applications. The techniques of Natural Language Processing (NLP) have shown remarkable capabilities for various tasks, including named entity recognition, machine translation and dialogue systems. Recently, Large Language Models (LLMs) have achieved significant improvements across various NLP downstream tasks. However, there is a lack of specialized LLMs for IT operations. In this paper, we introduce the OWL, a large language model trained on our collected OWL-Instruct dataset with a wide range of IT-related information, where the mixture-of-adapter strategy is proposed to improve the parameter-efficient tuning across different domains or tasks. Furthermore, we evaluate the performance of our OWL on the OWL-Bench established by us and open IT-related benchmarks. OWL demonstrates superior performance results on IT tasks, which outperforms existing models by significant margins. Moreover, we hope that the findings of our work will provide more insights to revolutionize the techniques of IT operations with specialized LLMs.

Yutong Qu, Wei Emma Zhang, Jian Yang et al.

Knowledge-aware methods have boosted a range of natural language processing applications over the last decades. With the gathered momentum, knowledge recently has been pumped into enormous attention in document summarization, one of natural language processing applications. Previous works reported that knowledge-embedded document summarizers excel at generating superior digests, especially in terms of informativeness, coherence, and fact consistency. This paper pursues to present the first systematic survey for the state-of-the-art methodologies that embed knowledge into document summarizers. Particularly, we propose novel taxonomies to recapitulate knowledge and knowledge embeddings under the document summarization view. We further explore how embeddings are generated in embedding learning architectures of document summarization models, especially of deep learning models. At last, we discuss the challenges of this topic and future directions.

Chuang Zhang, Li Shen, Jian Yang et al.

The memorization effect of deep neural networks (DNNs) plays a pivotal role in recent label noise learning methods. To exploit this effect, the model prediction-based methods have been widely adopted, which aim to exploit the outputs of DNNs in the early stage of learning to correct noisy labels. However, we observe that the model will make mistakes during label prediction, resulting in unsatisfactory performance. By contrast, the produced features in the early stage of learning show better robustness. Inspired by this observation, in this paper, we propose a novel feature embedding-based method for deep learning with label noise, termed LabEl NoiseDilution (LEND). To be specific, we first compute a similarity matrix based on current embedded features to capture the local structure of training data. Then, the noisy supervision signals carried by mislabeled data are overwhelmed by nearby correctly labeled ones (\textit{i.e.}, label noise dilution), of which the effectiveness is guaranteed by the inherent robustness of feature embedding. Finally, the training data with diluted labels are further used to train a robust classifier. Empirically, we conduct extensive experiments on both synthetic and real-world noisy datasets by comparing our LEND with several representative robust learning approaches. The results verify the effectiveness of our LEND.

Haobo Jiang, Jin Xie, Jian Yang

Double Q-learning is a popular reinforcement learning algorithm in Markov decision process (MDP) problems. Clipped Double Q-learning, as an effective variant of Double Q-learning, employs the clipped double estimator to approximate the maximum expected action value. Due to the underestimation bias of the clipped double estimator, the performance of clipped Double Q-learning may be degraded in some stochastic environments. In this paper, in order to reduce the underestimation bias, we propose an action candidate-based clipped double estimator for Double Q-learning. Specifically, we first select a set of elite action candidates with high action values from one set of estimators. Then, among these candidates, we choose the highest valued action from the other set of estimators. Finally, we use the maximum value in the second set of estimators to clip the action value of the chosen action in the first set of estimators and the clipped value is used for approximating the maximum expected action value. Theoretically, the underestimation bias in our clipped Double Q-learning decays monotonically as the number of action candidates decreases. Moreover, the number of action candidates controls the trade-off between the overestimation and underestimation biases. In addition, we also extend our clipped Double Q-learning to continuous action tasks via approximating the elite continuous action candidates. We empirically verify that our algorithm can more accurately estimate the maximum expected action value on some toy environments and yield good performance on several benchmark problems.

Hui Lv, Zhen Cui, Biao Wang et al.

Anomaly identification is highly dependent on the relationship between the object and the scene, as different/same object actions in same/different scenes may lead to various degrees of normality and anomaly. Therefore, object-scene relation actually plays a crucial role in anomaly detection but is inadequately explored in previous works. In this paper, we propose a Spatial-Temporal Relation Learning (STRL) framework to tackle the video anomaly detection task. First, considering dynamic characteristics of the objects as well as scene areas, we construct a Spatio-Temporal Auto-Encoder (STAE) to jointly exploit spatial and temporal evolution patterns for representation learning. For better pattern extraction, two decoding branches are designed in the STAE module, i.e. an appearance branch capturing spatial cues by directly predicting the next frame, and a motion branch focusing on modeling the dynamics via optical flow prediction. Then, to well concretize the object-scene relation, a Relation Learning (RL) module is devised to analyze and summarize the normal relations by introducing the Knowledge Graph Embedding methodology. Specifically in this process, the plausibility of object-scene relation is measured by jointly modeling object/scene features and optimizable object-scene relation maps. Extensive experiments are conducted on three public datasets, and the superior performance over the state-of-the-art methods demonstrates the effectiveness of our method.

Bing Wang, Changyu Ren, Jian Yang et al.

Recent LLM-based Text-to-SQL methods usually suffer from significant performance degradation on "huge" databases and complex user questions that require multi-step reasoning. Moreover, most existing methods neglect the crucial significance of LLMs utilizing external tools and model collaboration. To address these challenges, we introduce MAC-SQL, a novel LLM-based multi-agent collaborative framework. Our framework comprises a core decomposer agent for Text-to-SQL generation with few-shot chain-of-thought reasoning, accompanied by two auxiliary agents that utilize external tools or models to acquire smaller sub-databases and refine erroneous SQL queries. The decomposer agent collaborates with auxiliary agents, which are activated as needed and can be expanded to accommodate new features or tools for effective Text-to-SQL parsing. In our framework, We initially leverage GPT-4 as the strong backbone LLM for all agent tasks to determine the upper bound of our framework. We then fine-tune an open-sourced instruction-followed model, SQL-Llama, by leveraging Code Llama 7B, to accomplish all tasks as GPT-4 does. Experiments show that SQL-Llama achieves a comparable execution accuracy of 43.94, compared to the baseline accuracy of 46.35 for vanilla GPT-4. At the time of writing, MAC-SQL+GPT-4 achieves an execution accuracy of 59.59 when evaluated on the BIRD benchmark, establishing a new state-of-the-art (SOTA) on its holdout test set (https://github.com/wbbeyourself/MAC-SQL).

Zixiang Wang, Jian Yang, Tongliang Li et al.

Entity and relation extraction is a key task in information extraction, where the output can be used for downstream NLP tasks. Existing approaches for entity and relation extraction tasks mainly focus on the English corpora and ignore other languages. Thus, it is critical to improving performance in a multilingual setting. Meanwhile, multilingual training is usually used to boost cross-lingual performance by transferring knowledge from languages (e.g., high-resource) to other (e.g., low-resource) languages. However, language interference usually exists in multilingual tasks as the model parameters are shared among all languages. In this paper, we propose a two-stage multilingual training method and a joint model called Multilingual Entity and Relation Extraction framework (mERE) to mitigate language interference across languages. Specifically, we randomly concatenate sentences in different languages to train a Language-universal Aggregator (LA), which narrows the distance of embedding representations by obtaining the unified language representation. Then, we separate parameters to mitigate interference via tuning a Language-specific Switcher (LS), which includes several independent sub-modules to refine the language-specific feature representation. After that, to enhance the relational triple extraction, the sentence representations concatenated with the relation feature are used to recognize the entities. Extensive experimental results show that our method outperforms both the monolingual and multilingual baseline methods. Besides, we also perform detailed analysis to show that mERE is lightweight but effective on relational triple extraction and mERE{} is easy to transfer to other backbone models of multi-field tasks, which further demonstrates the effectiveness of our method.

Venus Haghighi, Behnaz Soltani, Adnan Mahmood et al.

Anomaly detection in attributed networks has received a considerable attention in recent years due to its applications in a wide range of domains such as finance, network security, and medicine. Traditional approaches cannot be adopted on attributed networks' settings to solve the problem of anomaly detection. The main limitation of such approaches is that they inherently ignore the relational information between data features. With a rapid explosion in deep learning- and graph neural networks-based techniques, spotting rare objects on attributed networks has significantly stepped forward owing to the potentials of deep techniques in extracting complex relationships. In this paper, we propose a new architecture on anomaly detection. The main goal of designing such an architecture is to utilize multi-task learning which would enhance the detection performance. Multi-task learning-based anomaly detection is still in its infancy and only a few studies in the existing literature have catered to the same. We incorporate both community detection and multi-view representation learning techniques for extracting distinct and complementary information from attributed networks and subsequently fuse the captured information for achieving a better detection result. The mutual collaboration between two main components employed in this architecture, i.e., community-specific learning and multi-view representation learning, exhibits a promising solution to reach more effective results.

Jianzhang Zheng, Hao Shen, Jian Yang et al.

Autoencoders have achieved great success in various computer vision applications. The autoencoder learns appropriate low dimensional image representations through the self-supervised paradigm, i.e., reconstruction. Existing studies mainly focus on the minimizing the reconstruction error on pixel level of image, while ignoring the preservation of Intrinsic Dimension (ID), which is a fundamental geometric property of data representations in Deep Neural Networks (DNNs). Motivated by the important role of ID, in this paper, we propose a novel deep representation learning approach with autoencoder, which incorporates regularization of the global and local ID constraints into the reconstruction of data representations. This approach not only preserves the global manifold structure of the whole dataset, but also maintains the local manifold structure of the feature maps of each point, which makes the learned low-dimensional features more discriminant and improves the performance of the downstream algorithms. To our best knowledge, existing works are rare and limited on exploiting both global and local ID invariant properties on the regularization of autoencoders. Numerical experimental results on benchmark datasets (Extended Yale B, Caltech101 and ImageNet) show that the resulting regularized learning models achieve better discriminative representations for downstream tasks including image classification and clustering.

Deguang Kong, Miao Lu, Konstantin Shmakov et al.

Consensus clustering aggregates partitions in order to find a better fit by reconciling clustering results from different sources/executions. In practice, there exist noise and outliers in clustering task, which, however, may significantly degrade the performance. To address this issue, we propose a novel algorithm -- robust consensus clustering that can find common ground truth among experts' opinions, which tends to be minimally affected by the bias caused by the outliers. In particular, we formalize the robust consensus clustering problem as a constraint optimization problem, and then derive an effective algorithm upon alternating direction method of multipliers (ADMM) with rigorous convergence guarantee. Our method outperforms the baselines on benchmarks. We apply the proposed method to the real-world advertising campaign segmentation and forecasting tasks using the proposed consensus clustering results based on the similarity computed via Kolmogorov-Smirnov Statistics. The accurate clustering result is helpful for building the advertiser profiles so as to perform the forecasting.

Xin Zhang, Xue Yang, Yuxuan Li et al.

Rotated object detection has made significant progress in the optical remote sensing. However, advancements in the Synthetic Aperture Radar (SAR) field are laggard behind, primarily due to the absence of a large-scale dataset. Annotating such a dataset is inefficient and costly. A promising solution is to employ a weakly supervised model (e.g., trained with available horizontal boxes only) to generate pseudo-rotated boxes for reference before manual calibration. Unfortunately, the existing weakly supervised models exhibit limited accuracy in predicting the object's angle. Previous works attempt to enhance angle prediction by using angle resolvers that decouple angles into cosine and sine encodings. In this work, we first reevaluate these resolvers from a unified perspective of dimension mapping and expose that they share the same shortcomings: these methods overlook the unit cycle constraint inherent in these encodings, easily leading to prediction biases. To address this issue, we propose the Unit Cycle Resolver, which incorporates a unit circle constraint loss to improve angle prediction accuracy. Our approach can effectively improve the performance of existing state-of-the-art weakly supervised methods and even surpasses fully supervised models on existing optical benchmarks (i.e., DOTA-v1.0 dataset). With the aid of UCR, we further annotate and introduce RSAR, the largest multi-class rotated SAR object detection dataset to date. Extensive experiments on both RSAR and optical datasets demonstrate that our UCR enhances angle prediction accuracy. Our dataset and code can be found at: https://github.com/zhasion/RSAR.

Zheng Li, Yibing Song, Ming-Ming Cheng et al.

Textual-based prompt learning methods primarily employ multiple learnable soft prompts and hard class tokens in a cascading manner as text inputs, aiming to align image and text (category) spaces for downstream tasks. However, current training is restricted to aligning images with predefined known categories and cannot be associated with unknown categories. In this work, we propose utilizing universal attributes as a bridge to enhance the alignment between images and unknown categories. Specifically, we introduce an Attribute-anchored Textual Prompt learning method for vision-language models, named ATPrompt. This approach expands the learning space of soft prompts from the original one-dimensional category level into the multi-dimensional attribute level by incorporating multiple attribute tokens into the learnable soft prompts. Through this modification, we transform the text prompt from a category-centric form to an attribute-category hybrid form. Additionally, we introduce a straightforward differentiable attribute search method to identify representative and suitable attributes for downstream tasks. As an easy-to-use plug-in technique, ATPrompt can seamlessly replace the existing basic prompt format in textual-based methods, providing general improvements at a negligible computational cost. Extensive experiments across 11 datasets validate the effectiveness of our method. Code is publicly available at https://github.com/zhengli97/ATPrompt.

Linzheng Chai, Jian Yang, Tao Sun et al. · tsinghua

Chain-of-thought (CoT) has emerged as a powerful technique to elicit reasoning in large language models and improve a variety of downstream tasks. CoT mainly demonstrates excellent performance in English, but its usage in low-resource languages is constrained due to poor language generalization. To bridge the gap among different languages, we propose a cross-lingual instruction fine-tuning framework (xCOT) to transfer knowledge from high-resource languages to low-resource languages. Specifically, the multilingual instruction training data (xCOT-INSTRUCT) is created to encourage the semantic alignment of multiple languages. We introduce cross-lingual in-context few-shot learning (xICL)) to accelerate multilingual agreement in instruction tuning, where some fragments of source languages in examples are randomly substituted by their counterpart translations of target languages. During multilingual instruction tuning, we adopt the randomly online CoT strategy to enhance the multilingual reasoning ability of the large language model by first translating the query to another language and then answering in English. To further facilitate the language transfer, we leverage the high-resource CoT to supervise the training of low-resource languages with cross-lingual distillation. Experimental results on previous benchmarks demonstrate the superior performance of xCoT in reducing the gap among different languages, highlighting its potential to reduce the cross-lingual gap.

Penghai Zhao, Xin Zhang, Jiayue Cao et al.

The rapid growth of research in Pattern Analysis and Machine Intelligence (PAMI) has rendered literature reviews essential for consolidating and interpreting knowledge across its many subfields. In this work, we present a comprehensive tertiary analysis of PAMI reviews along three complementary dimensions: (i) identifying structural and statistical regularities in existing surveys; (ii) developing quantitative strategies that help researchers navigate and prioritize within the expanding review corpus; and (iii) critically assessing emerging AI-generated review systems. To support this study, we construct RiPAMI, a large-scale database containing more than 3,000 review articles, and combine narrative synthesis with statistical analysis to capture structural and content-level features. Our analyses reveal distinctive organizational patterns as well as persistent gaps in current review practices. Building on these insights, we propose practical, article-level strategies for indicator-guided navigation that move beyond simple citation counts. Finally, our evaluation of state-of-the-art AI-generated reviews indicates encouraging advances in coherence and organization, yet also highlights enduring weaknesses in reference retrieval, coverage of recent work, and the incorporation of visual elements. Together, these findings provide both a critical appraisal of existing review practices and a forward-looking perspective on how AI-generated reviews can evolve into trustworthy, customizable, and transformative complements to traditional human-authored surveys.

Runqiang Zang, Hongcheng Guo, Jian Yang et al.

In spite of the rapid advancements in unsupervised log anomaly detection techniques, the current mainstream models still necessitate specific training for individual system datasets, resulting in costly procedures and limited scalability due to dataset size, thereby leading to performance bottlenecks. Furthermore, numerous models lack cognitive reasoning capabilities, posing challenges in direct transferability to similar systems for effective anomaly detection. Additionally, akin to reconstruction networks, these models often encounter the "identical shortcut" predicament, wherein the majority of system logs are classified as normal, erroneously predicting normal classes when confronted with rare anomaly logs due to reconstruction errors. To address the aforementioned issues, we propose MLAD, a novel anomaly detection model that incorporates semantic relational reasoning across multiple systems. Specifically, we employ Sentence-bert to capture the similarities between log sequences and convert them into highly-dimensional learnable semantic vectors. Subsequently, we revamp the formulas of the Attention layer to discern the significance of each keyword in the sequence and model the overall distribution of the multi-system dataset through appropriate vector space diffusion. Lastly, we employ a Gaussian mixture model to highlight the uncertainty of rare words pertaining to the "identical shortcut" problem, optimizing the vector space of the samples using the maximum expectation model. Experiments on three real-world datasets demonstrate the superiority of MLAD.

Yongdong chi, Hanqing Wang, Zonghan Yang et al.

Text-to-SQL transforms the user queries from natural language to executable SQL programs, enabling non-experts to interact with complex databases. Existing prompt-based methods craft meticulous text guidelines and examples to facilitate SQL generation, but their accuracy is hindered by the large semantic gap between the texts and the low-resource SQL programs. In this work, we propose Pi-SQL, which incorporates the high-resource Python program as a pivot to bridge between the natural language query and SQL program. In particular, Pi-SQL first generates Python programs that provide fine-grained step-by-step guidelines in their code blocks or comments, and then produces an SQL program following the guidance of each Python program. The final SQL program matches the reference Python program's query results and, through selection from candidates generated by different strategies, achieves superior execution speed, with a reward-based valid efficiency score up to 4.55 higher than the best-performing baseline. Extensive experiments demonstrate the effectiveness of Pi-SQL, which improves the execution accuracy of the best-performing baseline by up to 3.20.

M-A-P Team, Xinrun Du, Yifan Yao et al.

Large language models (LLMs) have demonstrated remarkable proficiency in mainstream academic disciplines such as mathematics, physics, and computer science. However, human knowledge encompasses over 200 specialized disciplines, far exceeding the scope of existing benchmarks. The capabilities of LLMs in many of these specialized fields-particularly in light industry, agriculture, and service-oriented disciplines-remain inadequately evaluated. To address this gap, we present SuperGPQA, a comprehensive benchmark that evaluates graduate-level knowledge and reasoning capabilities across 285 disciplines. Our benchmark employs a novel Human-LLM collaborative filtering mechanism to eliminate trivial or ambiguous questions through iterative refinement based on both LLM responses and expert feedback. Our experimental results reveal significant room for improvement in the performance of current state-of-the-art LLMs across diverse knowledge domains (e.g., the reasoning-focused model DeepSeek-R1 achieved the highest accuracy of 61.82% on SuperGPQA), highlighting the considerable gap between current model capabilities and artificial general intelligence. Additionally, we present comprehensive insights from our management of a large-scale annotation process, involving over 80 expert annotators and an interactive Human-LLM collaborative system, offering valuable methodological guidance for future research initiatives of comparable scope.

Shawn Gavin, Tuney Zheng, Jiaheng Liu et al.

The long-context capabilities of large language models (LLMs) have been a hot topic in recent years. To evaluate the performance of LLMs in different scenarios, various assessment benchmarks have emerged. However, as most of these benchmarks focus on identifying key information to answer questions, which mainly requires the retrieval ability of LLMs, these benchmarks can partially represent the reasoning performance of LLMs from large amounts of information. Meanwhile, although LLMs often claim to have context windows of 32k, 128k, 200k, or even longer, these benchmarks fail to reveal the actual supported length of these LLMs. To address these issues, we propose the LongIns benchmark dataset, a challenging long-context instruction-based exam for LLMs, which is built based on the existing instruction datasets. Specifically, in our LongIns, we introduce three evaluation settings: Global Instruction & Single Task (GIST), Local Instruction & Single Task (LIST), and Local Instruction & Multiple Tasks (LIMT). Based on LongIns, we perform comprehensive evaluations on existing LLMs and have the following important findings: (1). The top-performing GPT-4 with 128k context length performs poorly on the evaluation context window of 16k in our LongIns. (2). For the multi-hop reasoning ability of many existing LLMs, significant efforts are still needed under short context windows (less than 4k).

Tao Sun, Linzheng Chai, Jian Yang et al.

Intermediate reasoning or acting steps have successfully improved large language models (LLMs) for handling various downstream natural language processing (NLP) tasks. When applying LLMs for code generation, recent works mainly focus on directing the models to articulate intermediate natural-language reasoning steps, as in chain-of-thought (CoT) prompting, and then output code with the natural language or other structured intermediate steps. However, such output is not suitable for code translation or generation tasks since the standard CoT has different logical structures and forms of expression with the code. In this work, we introduce the universal code (UniCode) as the intermediate representation. It is a description of algorithm steps using a mix of conventions of programming languages, such as assignment operator, conditional operator, and loop. Hence, we collect an instruction dataset UniCoder-Instruct to train our model UniCoder on multi-task learning objectives. UniCoder-Instruct comprises natural-language questions, code solutions, and the corresponding universal code. The alignment between the intermediate universal code representation and the final code solution significantly improves the quality of the generated code. The experimental results demonstrate that UniCoder with the universal code significantly outperforms the previous prompting methods by a large margin, showcasing the effectiveness of the structural clues in pseudo-code.

Jiaqi Bai, Hongcheng Guo, Jiaheng Liu et al.

Retrieval-enhanced text generation has shown remarkable progress on knowledge-intensive language tasks, such as open-domain question answering and knowledge-enhanced dialogue generation, by leveraging passages retrieved from a large passage corpus for delivering a proper answer given the input query. However, the retrieved passages are not ideal for guiding answer generation because of the discrepancy between retrieval and generation, i.e., the candidate passages are all treated equally during the retrieval procedure without considering their potential to generate a proper answer. This discrepancy makes a passage retriever deliver a sub-optimal collection of candidate passages to generate the answer. In this paper, we propose the GeneRative Knowledge Improved Passage Ranking (GripRank) approach, addressing the above challenge by distilling knowledge from a generative passage estimator (GPE) to a passage ranker, where the GPE is a generative language model used to measure how likely the candidate passages can generate the proper answer. We realize the distillation procedure by teaching the passage ranker learning to rank the passages ordered by the GPE. Furthermore, we improve the distillation quality by devising a curriculum knowledge distillation mechanism, which allows the knowledge provided by the GPE can be progressively distilled to the ranker through an easy-to-hard curriculum, enabling the passage ranker to correctly recognize the provenance of the answer from many plausible candidates. We conduct extensive experiments on four datasets across three knowledge-intensive language tasks. Experimental results show advantages over the state-of-the-art methods for both passage ranking and answer generation on the KILT benchmark.

Juncheng Wan, Jian Yang, Shuming Ma et al.

While end-to-end neural machine translation (NMT) has achieved impressive progress, noisy input usually leads models to become fragile and unstable. Generating adversarial examples as the augmented data has been proved to be useful to alleviate this problem. Existing methods for adversarial example generation (AEG) are word-level or character-level, which ignore the ubiquitous phrase structure. In this paper, we propose a Phrase-level Adversarial Example Generation (PAEG) framework to enhance the robustness of the translation model. Our method further improves the gradient-based word-level AEG method by adopting a phrase-level substitution strategy. We verify our method on three benchmarks, including LDC Chinese-English, IWSLT14 German-English, and WMT14 English-German tasks. Experimental results demonstrate that our approach significantly improves translation performance and robustness to noise compared to previous strong baselines.

Xu Zhang, Jian Yang, Haoyang Huang et al.

Existing document-level neural machine translation (NMT) models have sufficiently explored different context settings to provide guidance for target generation. However, little attention is paid to inaugurate more diverse context for abundant context information. In this paper, we propose a Selective Memory-augmented Neural Document Translation model to deal with documents containing large hypothesis space of the context. Specifically, we retrieve similar bilingual sentence pairs from the training corpus to augment global context and then extend the two-stream attention model with selective mechanism to capture local context and diverse global contexts. This unified approach allows our model to be trained elegantly on three publicly document-level machine translation datasets and significantly outperforms previous document-level NMT models.

Jian Yang, Shuming Ma, Haoyang Huang et al.

This report describes Microsoft's machine translation systems for the WMT21 shared task on large-scale multilingual machine translation. We participated in all three evaluation tracks including Large Track and two Small Tracks where the former one is unconstrained and the latter two are fully constrained. Our model submissions to the shared task were initialized with DeltaLM\footnote{\url{https://aka.ms/deltalm}}, a generic pre-trained multilingual encoder-decoder model, and fine-tuned correspondingly with the vast collected parallel data and allowed data sources according to track settings, together with applying progressive learning and iterative back-translation approaches to further improve the performance. Our final submissions ranked first on three tracks in terms of the automatic evaluation metric.

Yang Yang, Hongchen Wei, Hengshu Zhu et al.

The task of image captioning aims to generate captions directly from images via the automatically learned cross-modal generator. To build a well-performing generator, existing approaches usually need a large number of described images, which requires a huge effects on manual labeling. However, in real-world applications, a more general scenario is that we only have limited amount of described images and a large number of undescribed images. Therefore, a resulting challenge is how to effectively combine the undescribed images into the learning of cross-modal generator. To solve this problem, we propose a novel image captioning method by exploiting the Cross-modal Prediction and Relation Consistency (CPRC), which aims to utilize the raw image input to constrain the generated sentence in the commonly semantic space. In detail, considering that the heterogeneous gap between modalities always leads to the supervision difficulty of using the global embedding directly, CPRC turns to transform both the raw image and corresponding generated sentence into the shared semantic space, and measure the generated sentence from two aspects: 1) Prediction consistency. CPRC utilizes the prediction of raw image as soft label to distill useful supervision for the generated sentence, rather than employing the traditional pseudo labeling; 2) Relation consistency. CPRC develops a novel relation consistency between augmented images and corresponding generated sentences to retain the important relational knowledge. In result, CPRC supervises the generated sentence from both the informativeness and representativeness perspectives, and can reasonably use the undescribed images to learn a more effective generator under the semi-supervised scenario.

Jian Yang, Xinyu Hu, Gang Xiao et al.

Pre-trained language models learn informative word representations on a large-scale text corpus through self-supervised learning, which has achieved promising performance in fields of natural language processing (NLP) after fine-tuning. These models, however, suffer from poor robustness and lack of interpretability. We refer to pre-trained language models with knowledge injection as knowledge-enhanced pre-trained language models (KEPLMs). These models demonstrate deep understanding and logical reasoning and introduce interpretability. In this survey, we provide a comprehensive overview of KEPLMs in NLP. We first discuss the advancements in pre-trained language models and knowledge representation learning. Then we systematically categorize existing KEPLMs from three different perspectives. Finally, we outline some potential directions of KEPLMs for future research.

Yuchen Eleanor Jiang, Tianyu Liu, Shuming Ma et al.

Standard automatic metrics, e.g. BLEU, are not reliable for document-level MT evaluation. They can neither distinguish document-level improvements in translation quality from sentence-level ones, nor identify the discourse phenomena that cause context-agnostic translations. This paper introduces a novel automatic metric BlonDe to widen the scope of automatic MT evaluation from sentence to document level. BlonDe takes discourse coherence into consideration by categorizing discourse-related spans and calculating the similarity-based F1 measure of categorized spans. We conduct extensive comparisons on a newly constructed dataset BWB. The experimental results show that BlonDe possesses better selectivity and interpretability at the document-level, and is more sensitive to document-level nuances. In a large-scale human study, BlonDe also achieves significantly higher Pearson's r correlation with human judgments compared to previous metrics.

Shuming Ma, Jian Yang, Haoyang Huang et al.

Multilingual machine translation enables a single model to translate between different languages. Most existing multilingual machine translation systems adopt a randomly initialized Transformer backbone. In this work, inspired by the recent success of language model pre-training, we present XLM-T, which initializes the model with an off-the-shelf pretrained cross-lingual Transformer encoder and fine-tunes it with multilingual parallel data. This simple method achieves significant improvements on a WMT dataset with 10 language pairs and the OPUS-100 corpus with 94 pairs. Surprisingly, the method is also effective even upon the strong baseline with back-translation. Moreover, extensive analysis of XLM-T on unsupervised syntactic parsing, word alignment, and multilingual classification explains its effectiveness for machine translation. The code will be at https://aka.ms/xlm-t.

Zhuo Huang, Ying Tai, Chengjie Wang et al.

Semi-Supervised Learning (SSL) with mismatched classes deals with the problem that the classes-of-interests in the limited labeled data is only a subset of the classes in massive unlabeled data. As a result, the classes only possessed by the unlabeled data may mislead the classifier training and thus hindering the realistic landing of various SSL methods. To solve this problem, existing methods usually divide unlabeled data to in-distribution (ID) data and out-of-distribution (OOD) data, and directly discard or weaken the OOD data to avoid their adverse impact. In other words, they treat OOD data as completely useless and thus the potential valuable information for classification contained by them is totally ignored. To remedy this defect, this paper proposes a "Transferable OOD data Recycling" (TOOR) method which properly utilizes ID data as well as the "recyclable" OOD data to enrich the information for conducting class-mismatched SSL. Specifically, TOOR firstly attributes all unlabeled data to ID data or OOD data, among which the ID data are directly used for training. Then we treat the OOD data that have a close relationship with ID data and labeled data as recyclable, and employ adversarial domain adaptation to project them to the space of ID data and labeled data. In other words, the recyclability of an OOD datum is evaluated by its transferability, and the recyclable OOD data are transferred so that they are compatible with the distribution of known classes-of-interests. Consequently, our TOOR method extracts more information from unlabeled data than existing approaches, so it can achieve the improved performance which is demonstrated by the experiments on typical benchmark datasets.

Min Fu, Jiwei Guan, Xi Zheng et al.

Efficient and appropriate online customer service is essential to large e-commerce businesses. Existing solution recommendation methods for online customer service are unable to determine the best solutions at runtime, leading to poor satisfaction of end customers. This paper proposes a novel intelligent framework, called ICS-Assist, to recommend suitable customer service solutions for service staff at runtime. Specifically, we develop a generalizable two-stage machine learning model to identify customer service scenarios and determine customer service solutions based on a scenario-solution mapping table. We implement ICS-Assist and evaluate it using an over 6-month field study with Alibaba Group. In our experiment, over 12,000 customer service staff use ICS-Assist to serve for over 230,000 cases per day on average. The experimen-tal results show that ICS-Assist significantly outperforms the traditional manual method, and improves the solution acceptance rate, the solution coverage rate, the average service time, the customer satisfaction rate, and the business domain catering rate by up to 16%, 25%, 6%, 14% and 17% respectively, compared to the state-of-the-art methods.

Xueya Zhang, Tong Zhang, Wenting Zhao et al.

Graph convolutional networks (GCNs) have shown the powerful ability in text structure representation and effectively facilitate the task of text classification. However, challenges still exist in adapting GCN on learning discriminative features from texts due to the main issue of graph variants incurred by the textual complexity and diversity. In this paper, we propose a dual-attention GCN to model the structural information of various texts as well as tackle the graph-invariant problem through embedding two types of attention mechanisms, i.e. the connection-attention and hop-attention, into the classic GCN. To encode various connection patterns between neighbour words, connection-attention adaptively imposes different weights specified to neighbourhoods of each word, which captures the short-term dependencies. On the other hand, the hop-attention applies scaled coefficients to different scopes during the graph diffusion process to make the model learn more about the distribution of context, which captures long-term semantics in an adaptive way. Extensive experiments are conducted on five widely used datasets to evaluate our dual-attention GCN, and the achieved state-of-the-art performance verifies the effectiveness of dual-attention mechanisms.

Zixiang Ding, Rui Xia, Jianfei Yu et al.

Deep neural networks have recently been shown to achieve highly competitive performance in many computer vision tasks due to their abilities of exploring in a much larger hypothesis space. However, since most deep architectures like stacked RNNs tend to suffer from the vanishing-gradient and overfitting problems, their effects are still understudied in many NLP tasks. Inspired by this, we propose a novel multi-layer RNN model called densely connected bidirectional long short-term memory (DC-Bi-LSTM) in this paper, which essentially represents each layer by the concatenation of its hidden state and all preceding layers' hidden states, followed by recursively passing each layer's representation to all subsequent layers. We evaluate our proposed model on five benchmark datasets of sentence classification. DC-Bi-LSTM with depth up to 20 can be successfully trained and obtain significant improvements over the traditional Bi-LSTM with the same or even less parameters. Moreover, our model has promising performance compared with the state-of-the-art approaches.

Xiang Li, Tao Qin, Jian Yang et al.

Recurrent neural networks (RNNs) have achieved state-of-the-art performances in many natural language processing tasks, such as language modeling and machine translation. However, when the vocabulary is large, the RNN model will become very big (e.g., possibly beyond the memory capacity of a GPU device) and its training will become very inefficient. In this work, we propose a novel technique to tackle this challenge. The key idea is to use 2-Component (2C) shared embedding for word representations. We allocate every word in the vocabulary into a table, each row of which is associated with a vector, and each column associated with another vector. Depending on its position in the table, a word is jointly represented by two components: a row vector and a column vector. Since the words in the same row share the row vector and the words in the same column share the column vector, we only need $2 \sqrt{|V|}$ vectors to represent a vocabulary of $|V|$ unique words, which are far less than the $|V|$ vectors required by existing approaches. Based on the 2-Component shared embedding, we design a new RNN algorithm and evaluate it using the language modeling task on several benchmark datasets. The results show that our algorithm significantly reduces the model size and speeds up the training process, without sacrifice of accuracy (it achieves similar, if not better, perplexity as compared to state-of-the-art language models). Remarkably, on the One-Billion-Word benchmark Dataset, our algorithm achieves comparable perplexity to previous language models, whilst reducing the model size by a factor of 40-100, and speeding up the training process by a factor of 2. We name our proposed algorithm \emph{LightRNN} to reflect its very small model size and very high training speed.