Search

This paper proposes a new method, OFA-OCR, to transfer multimodal pretrained models to text recognition. Specifically, we recast text recognition as image captioning and directly transfer a unified vision-language pretrained model to the end task. Without pretraining on large-scale annotated or synthetic text recognition data, OFA-OCR outperforms the baselines and achieves state-of-the-art performance in the Chinese text recognition benchmark. Additionally, we construct an OCR pipeline with OFA-OCR, and we demonstrate that it can achieve competitive performance with the product-level API. The code (https://github.com/OFA-Sys/OFA) and demo (https://modelscope.cn/studios/damo/ofa_ocr_pipeline/summary) are publicly available.

Recently, attention based models have been used extensively in many sequence-to-sequence learning systems. Especially for image captioning, the attention based models are expected to ground correct image regions with proper generated words. However, for each time step in the decoding process, the attention based models usually use the hidden state of the current input to attend to the image regions. Under this setting, these attention models have a "deviated focus" problem that they calculate the attention weights based on previous words instead of the one to be generated, impairing the performance of both grounding and captioning. In this paper, we propose the Prophet Attention, similar to the form of self-supervision. In the training stage, this module utilizes the future information to calculate the "ideal" attention weights towards image regions. These calculated "ideal" weights are further used to regularize the "deviated" attention. In this manner, image regions are grounded with the correct words. The proposed Prophet Attention can be easily incorporated into existing image captioning models to improve their performance of both grounding and captioning. The experiments on the Flickr30k Entities and the MSCOCO datasets show that the proposed Prophet Attention consistently outperforms baselines in both automatic metrics and human evaluations. It is worth noticing that we set new state-of-the-arts on the two benchmark datasets and achieve the 1st place on the leaderboard of the online MSCOCO benchmark in terms of the default ranking score, i.e., CIDEr-c40., Accepted by NeurIPS 2020

Collaborative learning has successfully applied knowledge transfer to guide a pool of small student networks towards robust local minima. However, previous approaches typically struggle with drastically aggravated student homogenization when the number of students rises. In this paper, we propose Collaborative Group Learning, an efficient framework that aims to diversify the feature representation and conduct an effective regularization. Intuitively, similar to the human group study mechanism, we induce students to learn and exchange different parts of course knowledge as collaborative groups. First, each student is established by randomly routing on a modular neural network, which facilitates flexible knowledge communication between students due to random levels of representation sharing and branching. Second, to resist the student homogenization, students first compose diverse feature sets by exploiting the inductive bias from sub-sets of training data, and then aggregate and distill different complementary knowledge by imitating a random sub-group of students at each time step. Overall, the above mechanisms are beneficial for maximizing the student population to further improve the model generalization without sacrificing computational efficiency. Empirical evaluations on both image and text tasks indicate that our method significantly outperforms various state-of-the-art collaborative approaches whilst enhancing computational efficiency., Comment: Accepted by AAAI 2021; Camera ready version

Contrastive Language-Image Pre-training (CLIP) formulates image classification as an image-to-text matching task, i.e., matching images to the corresponding natural language descriptions instead of discrete category IDs. This allows for open-vocabulary visual recognition, where the model can recognize images from an open class set (also known as an open vocabulary) in a zero-shot manner. However, evaluating the openness of CLIP-like models is challenging, as the models are open to arbitrary vocabulary in theory, but their accuracy varies in practice. To address this, we resort to an incremental perspective to assess the openness through vocabulary expansions, and define extensibility to measure a model's ability to handle novel classes. Our evaluation shows that CLIP-like models are not truly open, and their performance deteriorates as the vocabulary expands. We further dissect the feature space of CLIP from the perspectives of representation alignment and uniformity. Our investigation reveals that the overestimation of openness is due to confusion among competing text features, rather than a failure to capture the similarity between image features and text features of novel classes. We hope that our investigation and analysis will facilitate future research on the CLIP openness issue., Accepted by Findings of ACL 2023 (Long Paper). Code is available at https://github.com/lancopku/clip-openness

Generalist models, which are capable of performing diverse multi-modal tasks in a task-agnostic way within a single model, have been explored recently. Being, hopefully, an alternative to approaching general-purpose AI, existing generalist models are still at an early stage, where modality and task coverage is limited. To empower multi-modal task-scaling and speed up this line of research, we release a generalist model learning system, OFASys, built on top of a declarative task interface named multi-modal instruction. At the core of OFASys is the idea of decoupling multi-modal task representations from the underlying model implementations. In OFASys, a task involving multiple modalities can be defined declaratively even with just a single line of code. The system automatically generates task plans from such instructions for training and inference. It also facilitates multi-task training for diverse multi-modal workloads. As a starting point, we provide presets of 7 different modalities and 23 highly-diverse example tasks in OFASys, with which we also develop a first-in-kind, single model, OFA+, that can handle text, image, speech, video, and motion data. The single OFA+ model achieves 95% performance in average with only 16% parameters of 15 task-finetuned models, showcasing the performance reliability of multi-modal task-scaling provided by OFASys. Available at https://github.com/OFA-Sys/OFASys

Investigating better ways to reuse the released pre-trained language models (PLMs) can significantly reduce the computational cost and the potential environmental side-effects. This paper explores a novel PLM reuse paradigm, Knowledge Integration (KI). Without human annotations available, KI aims to merge the knowledge from different teacher-PLMs, each of which specializes in a different classification problem, into a versatile student model. To achieve this, we first derive the correlation between virtual golden supervision and teacher predictions. We then design a Model Uncertainty--aware Knowledge Integration (MUKI) framework to recover the golden supervision for the student. Specifically, MUKI adopts Monte-Carlo Dropout to estimate model uncertainty for the supervision integration. An instance-wise re-weighting mechanism based on the margin of uncertainty scores is further incorporated, to deal with the potential conflicting supervision from teachers. Experimental results demonstrate that MUKI achieves substantial improvements over baselines on benchmark datasets. Further analysis shows that MUKI can generalize well for merging teacher models with heterogeneous architectures, and even teachers major in cross-lingual datasets., Comment: EMNLP 2022 (Findings), an improved version of arXiv:2112.07327. Code will be available at https://github.com/lancopku/MUKI

As many fine-tuned pre-trained language models~(PLMs) with promising performance are generously released, investigating better ways to reuse these models is vital as it can greatly reduce the retraining computational cost and the potential environmental side-effects. In this paper, we explore a novel model reuse paradigm, Knowledge Amalgamation~(KA) for PLMs. Without human annotations available, KA aims to merge the knowledge from different teacher-PLMs, each of which specializes in a different classification problem, into a versatile student model. The achieve this, we design a Model Uncertainty--aware Knowledge Amalgamation~(MUKA) framework, which identifies the potential adequate teacher using Monte-Carlo Dropout for approximating the golden supervision to guide the student. Experimental results demonstrate that MUKA achieves substantial improvements over baselines on benchmark datasets. Further analysis shows that MUKA can generalize well under several complicate settings with multiple teacher models, heterogeneous teachers, and even cross-dataset teachers.

The class imbalance problem, as an important issue in learning node representations, has drawn increasing attention from the community. Although the imbalance considered by existing studies roots from the unequal quantity of labeled examples in different classes (quantity imbalance), we argue that graph data expose a unique source of imbalance from the asymmetric topological properties of the labeled nodes, i.e., labeled nodes are not equal in terms of their structural role in the graph (topology imbalance). In this work, we first probe the previously unknown topology-imbalance issue, including its characteristics, causes, and threats to semi-supervised node classification learning. We then provide a unified view to jointly analyzing the quantity- and topology- imbalance issues by considering the node influence shift phenomenon with the Label Propagation algorithm. In light of our analysis, we devise an influence conflict detection -- based metric Totoro to measure the degree of graph topology imbalance and propose a model-agnostic method ReNode to address the topology-imbalance issue by re-weighting the influence of labeled nodes adaptively based on their relative positions to class boundaries. Systematic experiments demonstrate the effectiveness and generalizability of our method in relieving topology-imbalance issue and promoting semi-supervised node classification. The further analysis unveils varied sensitivity of different graph neural networks (GNNs) to topology imbalance, which may serve as a new perspective in evaluating GNN architectures., Accepted By NeurIPS 2021

Video captioning combines video understanding and language generation. Different from image captioning that describes a static image with details of almost every object, video captioning usually considers a sequence of frames and biases towards focused objects, e.g., the objects that stay in focus regardless of the changing background. Therefore, detecting and properly accommodating focused objects is critical in video captioning. To enforce the description of focused objects and achieve controllable video captioning, we propose an Object-Oriented Non-Autoregressive approach (O2NA), which performs caption generation in three steps: 1) identify the focused objects and predict their locations in the target caption; 2) generate the related attribute words and relation words of these focused objects to form a draft caption; and 3) combine video information to refine the draft caption to a fluent final caption. Since the focused objects are generated and located ahead of other words, it is difficult to apply the word-by-word autoregressive generation process; instead, we adopt a non-autoregressive approach. The experiments on two benchmark datasets, i.e., MSR-VTT and MSVD, demonstrate the effectiveness of O2NA, which achieves results competitive with the state-of-the-arts but with both higher diversity and higher inference speed., Accepted by Findings of ACL 2021 (The Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing)

Skip connection, is a widely-used technique to improve the performance and the convergence of deep neural networks, which is believed to relieve the difficulty in optimization due to non-linearity by propagating a linear component through the neural network layers. However, from another point of view, it can also be seen as a modulating mechanism between the input and the output, with the input scaled by a pre-defined value one. In this work, we investigate how the scale factors in the effectiveness of the skip connection and reveal that a trivial adjustment of the scale will lead to spurious gradient exploding or vanishing in line with the deepness of the models, which could be addressed by normalization, in particular, layer normalization, which induces consistent improvements over the plain skip connection. Inspired by the findings, we further propose to adaptively adjust the scale of the input by recursively applying skip connection with layer normalization, which promotes the performance substantially and generalizes well across diverse tasks including both machine translation and image classification datasets., Comment: Accepted by COLING2020 (The 28th International Conference on Computational Linguistics (COLING 2020))

Graph Contrastive Learning (GCL) has proven highly effective in promoting the performance of Semi-Supervised Node Classification (SSNC). However, existing GCL methods are generally transferred from other fields like CV or NLP, whose underlying working mechanism remains under-explored. In this work, we first deeply probe the working mechanism of GCL in SSNC, and find that the promotion brought by GCL is severely unevenly distributed: the improvement mainly comes from subgraphs with less annotated information, which is fundamentally different from contrastive learning in other fields. However, existing GCL methods generally ignore this uneven distribution of annotated information and apply GCL evenly to the whole graph. To remedy this issue and further improve GCL in SSNC, we propose the Topology InFormation gain-Aware Graph Contrastive Learning (TIFA-GCL) framework that considers the annotated information distribution across graph in GCL. Extensive experiments on six benchmark graph datasets, including the enormous OGB-Products graph, show that TIFA-GCL can bring a larger improvement than existing GCL methods in both transductive and inductive settings. Further experiments demonstrate the generalizability and interpretability of TIFA-GCL., Accepted by IJCAI-2022

In sequence-to-sequence learning, e.g., natural language generation, the decoder relies on the attention mechanism to efficiently extract information from the encoder. While it is common practice to draw information from only the last encoder layer, recent work has proposed to use representations from different encoder layers for diversified levels of information. Nonetheless, the decoder still obtains only a single view of the source sequences, which might lead to insufficient training of the encoder layer stack due to the hierarchy bypassing problem. In this work, we propose layer-wise multi-view decoding, where for each decoder layer, together with the representations from the last encoder layer, which serve as a global view, those from other encoder layers are supplemented for a stereoscopic view of the source sequences. Systematic experiments and analyses show that we successfully address the hierarchy bypassing problem, require almost negligible parameter increase, and substantially improve the performance of sequence-to-sequence learning with deep representations on five diverse tasks, i.e., machine translation, abstractive summarization, image captioning, video captioning, medical report generation, and paraphrase generation. In particular, our approach achieves new state-of-the-art results on ten benchmark datasets, including a low-resource machine translation dataset and two low-resource medical report generation datasets.

Pre-trained self-supervised models such as BERT have achieved striking success in learning sequence representations, especially for natural language processing. These models typically corrupt the given sequences with certain types of noise, such as masking, shuffling, or substitution, and then try to recover the original input. However, such pre-training approaches are prone to learning representations that are covariant with the noise, leading to the discrepancy between the pre-training and fine-tuning stage. To remedy this, we present ContrAstive Pre-Training (CAPT) to learn noise invariant sequence representations. The proposed CAPT encourages the consistency between representations of the original sequence and its corrupted version via unsupervised instance-wise training signals. In this way, it not only alleviates the pretrain-finetune discrepancy induced by the noise of pre-training, but also aids the pre-trained model in better capturing global semantics of the input via more effective sentence-level supervision. Different from most prior work that focuses on a particular modality, comprehensive empirical evidence on 11 natural language understanding and cross-modal tasks illustrates that CAPT is applicable for both language and vision-language tasks, and obtains surprisingly consistent improvement, including 0.6\% absolute gain on GLUE benchmarks and 0.8\% absolute increment on $\text{NLVR}^2$., Comment: Corrected typos

Training deep neural networks requires intricate initialization and careful selection of learning rates. The emergence of stochastic gradient optimization methods that use adaptive learning rates based on squared past gradients, e.g., AdaGrad, AdaDelta, and Adam, eases the job slightly. However, such methods have also been proven problematic in recent studies with their own pitfalls including non-convergence issues and so on. Alternative variants have been proposed for enhancement, such as AMSGrad, AdaShift and AdaBound. In this work, we identify a new problem of adaptive learning rate methods that exhibits at the beginning of learning where Adam produces extremely large learning rates that inhibit the start of learning. We propose the Adaptive and Momental Bound (AdaMod) method to restrict the adaptive learning rates with adaptive and momental upper bounds. The dynamic learning rate bounds are based on the exponential moving averages of the adaptive learning rates themselves, which smooth out unexpected large learning rates and stabilize the training of deep neural networks. Our experiments verify that AdaMod eliminates the extremely large learning rates throughout the training and brings significant improvements especially on complex networks such as DenseNet and Transformer, compared to Adam. Our implementation is available at: https://github.com/lancopku/AdaMod

Chinese word segmentation (CWS) is a fundamental step of Chinese natural language processing. In this paper, we build a new toolkit, named PKUSEG, for multi-domain word segmentation. Unlike existing single-model toolkits, PKUSEG targets multi-domain word segmentation and provides separate models for different domains, such as web, medicine, and tourism. Besides, due to the lack of labeled data in many domains, we propose a domain adaptation paradigm to introduce cross-domain semantic knowledge via a translation system. Through this method, we generate synthetic data using a large amount of unlabeled data in the target domain and then obtain a word segmentation model for the target domain. We also further refine the performance of the default model with the help of synthetic data. Experiments show that PKUSEG achieves high performance on multiple domains. The new toolkit also supports POS tagging and model training to adapt to various application scenarios. The toolkit is now freely and publicly available for the usage of research and industry.

In vision-and-language grounding problems, fine-grained representations of the image are considered to be of paramount importance. Most of the current systems incorporate visual features and textual concepts as a sketch of an image. However, plainly inferred representations are usually undesirable in that they are composed of separate components, the relations of which are elusive. In this work, we aim at representing an image with a set of integrated visual regions and corresponding textual concepts, reflecting certain semantics. To this end, we build the Mutual Iterative Attention (MIA) module, which integrates correlated visual features and textual concepts, respectively, by aligning the two modalities. We evaluate the proposed approach on two representative vision-and-language grounding tasks, i.e., image captioning and visual question answering. In both tasks, the semantic-grounded image representations consistently boost the performance of the baseline models under all metrics across the board. The results demonstrate that our approach is effective and generalizes well to a wide range of models for image-related applications. (The code is available at https://github.com/fenglinliu98/MIA), Accepted by NeurIPS 2019

Chinese word segmentation (CWS) is a fundamental step of Chinese natural language processing. In this paper, we build a new toolkit, named PKUSEG, for multi-domain word segmentation. Unlike existing single-model toolkits, PKUSEG targets at multi-domain word segmentation and provides separate models for different domains, such as web, medicine, and tourism. The new toolkit also supports POS tagging and model training to adapt to various application scenarios. Experiments show that PKUSEG achieves high performance on multiple domains. The toolkit is now freely and publicly available for the usage of research and industry.

Automatic evaluation of semantic rationality is an important yet challenging task, and current automatic techniques cannot well identify whether a sentence is semantically rational. The methods based on the language model do not measure the sentence by rationality but by commonness. The methods based on the similarity with human written sentences will fail if human-written references are not available. In this paper, we propose a novel model called Sememe-Word-Matching Neural Network (SWM-NN) to tackle semantic rationality evaluation by taking advantage of sememe knowledge base HowNet. The advantage is that our model can utilize a proper combination of sememes to represent the fine-grained semantic meanings of a word within the specific contexts. We use the fine-grained semantic representation to help the model learn the semantic dependency among words. To evaluate the effectiveness of the proposed model, we build a large-scale rationality evaluation dataset. Experimental results on this dataset show that the proposed model outperforms the competitive baselines with a 5.4\% improvement in accuracy.

Huge numbers of new words emerge every day, leading to a great need for representing them with semantic meaning that is understandable to NLP systems. Sememes are defined as the minimum semantic units of human languages, the combination of which can represent the meaning of a word. Manual construction of sememe based knowledge bases is time-consuming and labor-intensive. Fortunately, communities are devoted to composing the descriptions of words in the wiki websites. In this paper, we explore to automatically predict lexical sememes based on the descriptions of the words in the wiki websites. We view this problem as a weakly ordered multi-label task and propose a Label Distributed seq2seq model (LD-seq2seq) with a novel soft loss function to solve the problem. In the experiments, we take a real-world sememe knowledge base HowNet and the corresponding descriptions of the words in Baidu Wiki for training and evaluation. The results show that our LD-seq2seq model not only beats all the baselines significantly on the test set, but also outperforms amateur human annotators in a random subset of the test set.

This paper explores a new natural language processing task, review-driven multi-label music style classification. This task requires the system to identify multiple styles of music based on its reviews on websites. The biggest challenge lies in the complicated relations of music styles. It has brought failure to many multi-label classification methods. To tackle this problem, we propose a novel deep learning approach to automatically learn and exploit style correlations. The proposed method consists of two parts: a label-graph based neural network, and a soft training mechanism with correlation-based continuous label representation. Experimental results show that our approach achieves large improvements over the baselines on the proposed dataset. Especially, the micro F1 is improved from 53.9 to 64.5, and the one-error is reduced from 30.5 to 22.6. Furthermore, the visualized analysis shows that our approach performs well in capturing style correlations.

Existing text generation methods tend to produce repeated and "boring" expressions. To tackle this problem, we propose a new text generation model, called Diversity-Promoting Generative Adversarial Network (DP-GAN). The proposed model assigns low reward for repeatedly generated text and high reward for "novel" and fluent text, encouraging the generator to produce diverse and informative text. Moreover, we propose a novel language-model based discriminator, which can better distinguish novel text from repeated text without the saturation problem compared with existing classifier-based discriminators. The experimental results on review generation and dialogue generation tasks demonstrate that our model can generate substantially more diverse and informative text than existing baselines. The code is available at https://github.com/lancopku/DPGAN, Comment: Accepted by EMNLP 2018

A great proportion of sequence-to-sequence (Seq2Seq) models for Neural Machine Translation (NMT) adopt Recurrent Neural Network (RNN) to generate translation word by word following a sequential order. As the studies of linguistics have proved that language is not linear word sequence but sequence of complex structure, translation at each step should be conditioned on the whole target-side context. To tackle the problem, we propose a new NMT model that decodes the sequence with the guidance of its structural prediction of the context of the target sequence. Our model generates translation based on the structural prediction of the target-side context so that the translation can be freed from the bind of sequential order. Experimental results demonstrate that our model is more competitive compared with the state-of-the-art methods, and the analysis reflects that our model is also robust to translating sentences of different lengths and it also reduces repetition with the instruction from the target-side context for decoding., Comment: Accepted by COLING 2018

Web 2.0 has brought with it numerous user-produced data revealing one's thoughts, experiences, and knowledge, which are a great source for many tasks, such as information extraction, and knowledge base construction. However, the colloquial nature of the texts poses new challenges for current natural language processing techniques, which are more adapt to the formal form of the language. Ellipsis is a common linguistic phenomenon that some words are left out as they are understood from the context, especially in oral utterance, hindering the improvement of dependency parsing, which is of great importance for tasks relied on the meaning of the sentence. In order to promote research in this area, we are releasing a Chinese dependency treebank of 319 weibos, containing 572 sentences with omissions restored and contexts reserved., Comment: The treebank is available at https://github.com/lancopku/Chinese-Dependency-Treebank-with-Ellipsis

In the training of transition-based dependency parsers, an oracle is used to predict a transition sequence for a sentence and its gold tree. However, the transition system may exhibit ambiguity, that is, there can be multiple correct transition sequences that form the gold tree. We propose to make use of the property in the training of neural dependency parsers, and present the Hybrid Oracle. The new oracle gives all the correct transitions for a parsing state, which are used in the cross entropy loss function to provide better supervisory signal. It is also used to generate different transition sequences for a sentence to better explore the training data and improve the generalization ability of the parser. Evaluations show that the parsers trained using the hybrid oracle outperform the parsers using the traditional oracle in Chinese dependency parsing. We provide analysis from a linguistic view. The code is available at https://github.com/lancopku/nndep .

Existing neural models usually predict the tag of the current token independent of the neighboring tags. The popular LSTM-CRF model considers the tag dependencies between every two consecutive tags. However, it is hard for existing neural models to take longer distance dependencies of tags into consideration. The scalability is mainly limited by the complex model structures and the cost of dynamic programming during training. In our work, we first design a new model called "high order LSTM" to predict multiple tags for the current token which contains not only the current tag but also the previous several tags. We call the number of tags in one prediction as "order". Then we propose a new method called Multi-Order BiLSTM (MO-BiLSTM) which combines low order and high order LSTMs together. MO-BiLSTM keeps the scalability to high order models with a pruning technique. We evaluate MO-BiLSTM on all-phrase chunking and NER datasets. Experiment results show that MO-BiLSTM achieves the state-of-the-art result in chunking and highly competitive results in two NER datasets., Accepted by COLING 2018

As traditional neural network consumes a significant amount of computing resources during back propagation, \citet{Sun2017mePropSB} propose a simple yet effective technique to alleviate this problem. In this technique, only a small subset of the full gradients are computed to update the model parameters. In this paper we extend this technique into the Convolutional Neural Network(CNN) to reduce calculation in back propagation, and the surprising results verify its validity in CNN: only 5\% of the gradients are passed back but the model still achieves the same effect as the traditional CNN, or even better. We also show that the top-$k$ selection of gradients leads to a sparse calculation in back propagation, which may bring significant computational benefits for high computational complexity of convolution operation in CNN.

We propose a simple yet effective technique for neural network learning. The forward propagation is computed as usual. In back propagation, only a small subset of the full gradient is computed to update the model parameters. The gradient vectors are sparsified in such a way that only the top-$k$ elements (in terms of magnitude) are kept. As a result, only $k$ rows or columns (depending on the layout) of the weight matrix are modified, leading to a linear reduction ($k$ divided by the vector dimension) in the computational cost. Surprisingly, experimental results demonstrate that we can update only 1-4% of the weights at each back propagation pass. This does not result in a larger number of training iterations. More interestingly, the accuracy of the resulting models is actually improved rather than degraded, and a detailed analysis is given. The code is available at https://github.com/lancopku/meProp, Accepted by the 34th International Conference on Machine Learning (ICML 2017)

We propose a method, called Label Embedding Network, which can learn label representation (label embedding) during the training process of deep networks. With the proposed method, the label embedding is adaptively and automatically learned through back propagation. The original one-hot represented loss function is converted into a new loss function with soft distributions, such that the originally unrelated labels have continuous interactions with each other during the training process. As a result, the trained model can achieve substantially higher accuracy and with faster convergence speed. Experimental results based on competitive tasks demonstrate the effectiveness of the proposed method, and the learned label embedding is reasonable and interpretable. The proposed method achieves comparable or even better results than the state-of-the-art systems. The source code is available at \url{https://github.com/lancopku/LabelEmb}.

Recent systems on structured prediction focus on increasing the level of structural dependencies within the model. However, our study suggests that complex structures entail high overfitting risks. To control the structure-based overfitting, we propose to conduct structure regularization decoding (SR decoding). The decoding of the complex structure model is regularized by the additionally trained simple structure model. We theoretically analyze the quantitative relations between the structural complexity and the overfitting risk. The analysis shows that complex structure models are prone to the structure-based overfitting. Empirical evaluations show that the proposed method improves the performance of the complex structure models by reducing the structure-based overfitting. On the sequence labeling tasks, the proposed method substantially improves the performance of the complex neural network models. The maximum F1 error rate reduction is 36.4% for the third-order model. The proposed method also works for the parsing task. The maximum UAS improvement is 5.5% for the tri-sibling model. The results are competitive with or better than the state-of-the-art results., Comment: arXiv admin note: text overlap with arXiv:1411.6243