【摘 要】
：
Most State-Of-The-Art (SOTA) Neural Machine Translation (NMT) systems today achieve outstanding results based only on large parallel corpora.The large-scale parallel corpora for high-resource languages is easily obtainable.However,the translation quality
【机 构】
：
Institute for Artificial Intelligence,Beijing National Research Center for Information Science and T
论文部分内容阅读
Most State-Of-The-Art (SOTA) Neural Machine Translation (NMT) systems today achieve outstanding results based only on large parallel corpora.The large-scale parallel corpora for high-resource languages is easily obtainable.However,the translation quality of NMT for morphologically rich languages is still unsatisfactory,mainly because of the data sparsity problem encountered in Low-Resource Languages (LRLs).In the low-resource NMT paradigm,Transfer Learning (TL) has been developed into one of the most efficient methods.It is difficult to train the model on high-resource languages to include the information in both parent and child models,as well as the initially trained model that only contains the lexicon features and word embeddings of the parent model instead of the child languages feature.In this work,we aim to address this issue by proposing the language-independent Hybrid Transfer Learning (HTL) method for LRLs by sharing lexicon embedding between parent and child languages without leveraging back translation or manually injecting noises.First,we train the High-Resource Languages (HRLs) as the parent model with its vocabularies.Then,we combine the parent and child language pairs using the oversampling method to train the hybrid model initialized by the previously parent model.Finally,we fine-tune the morphologically rich child model using a hybrid model.Besides,we explore some exciting discoveries on the original TL approach.Experimental results show that our model consistently outperforms five SOTA methods in two languages Azerbaijani(Az) and Uzbek (Uz).Meanwhile,our approach is practical and significantly better,achieving improvements of up to 4.94 and 4.84 BLEU points for low-resource child languages Az → Zh and Uz → Zh,respectively.
其他文献
Event temporal relation extraction is an important part of natural language processing.Many models are being used in this task with the development of deep learning.However,most of the existing methods cannot accurately obtain the degree of association be
Road pricing is an urban traffic management mechanism to reduce traffic congestion.Currently,most of the road pricing systems based on predefined charging tolls fail to consider the dynamics of urban traffic flows and travelers\' demands on the arrival
Integer overflow is a common vulnerability in Ethereum Smart Contracts (ESCs) and often causes huge economic losses.Smart contracts cannot be changed once it is deployed on the blockchain and thus demand further testing.Mutation testing is a fault-based t
Lesion detection in Computed Tomography (CT) images is a challenging task in the field of computer-aided diagnosis.An important issue is to locate the area of lesion accurately.As a branch of Convolutional Neural Networks (CNNs),3D Context-Enhanced (3DCE)
In distributed training,increasing batch size can improve parallelism,but it can also bring many difficulties to the training process and cause training errors.In this work,we investigate the occurrence of training errors in theory and train ResNet-50 on
Identifying the association between metabolites and diseases will help us understand the pathogenesis of diseases,which has great significance in diagnosing and treating diseases.However,traditional biometric methods are time consuming and expensive.Accor
N400 is an objective electrophysiological index in semantic processing for brain.This study focuses on the sensitivity of N400 effect during speech comprehension under the uni-and bi-modality conditions.Varying the Signal-to-Noise Ratio (SNR) of speech si
The prevalence of missing values in the data streams collected in real environments makes them impossible to ignore in the privacy preservation of data streams.However,the development of most privacy preservation methods does not consider missing values.A
Online advertising click-through rate (CTR) prediction is aimed at predicting the probability of a user clicking an ad,and it has undergone considerable development in recent years.One of the hot topics in this area is the construction of feature interact
The proliferation of massive datasets has led to significant interests in distributed algorithms for solving large-scale machine learning problems.However,the communication overhead is a major bottleneck that hampers the scalability of distributed machine