Selective and Adaptive Incremental Transfer Learning with Multiple Datasets for Machine Fault Diagnosis

The visions of Industry 4.0 and 5.0 have reinforced the industrial environment.They have also made artificial intelligence incorporated as a major facilitator.Diagnosing machine faults has become a solid foundation for automatically recognizing machine failure,and thus timely maintenance can ensure safe operations.Transfer learning is a promising solution that can enhance the machine fault diagnosis model by borrowing pre-trained knowledge from the source model and applying it to the target model,which typically involves two datasets.In response to the availability of multiple datasets,this paper proposes using selective and adaptive incremental transfer learning(SA-ITL),which fuses three algorithms,namely,the hybrid selective algorithm,the transferability enhancement algorithm,and the incremental transfer learning algorithm.It is a selective algorithm that enables selecting and ordering appropriate datasets for transfer learning and selecting useful knowledge to avoid negative transfer.The algorithm also adaptively adjusts the portion of training data to balance the learning rate and training time.The proposed algorithm is evaluated and analyzed using ten benchmark datasets.Compared with other algorithms from existing works,SA-ITL improves the accuracy of all datasets.Ablation studies present the accuracy enhancements of the SA-ITL,including the hybrid selective algorithm(1.22%-3.82%),transferability enhancement algorithm(1.91%-4.15%),and incremental transfer learning algorithm(0.605%-2.68%).These also show the benefits of enhancing the target model with heterogeneous image datasets that widen the range of domain selection between source and target domains.

A Survey on Negative Transfer

Transfer learning(TL)utilizes data or knowledge from one or more source domains to facilitate learning in a target domain.It is particularly useful when the target domain has very few or no labeled data,due to annotation expense,privacy concerns,etc.Unfortunately,the effectiveness of TL is not always guaranteed.Negative transfer(NT),i.e.,leveraging source domain data/knowledge undesirably reduces learning performance in the target domain,and has been a long-standing and challenging problem in TL.Various approaches have been proposed in the literature to address this issue.However,there does not exist a systematic survey.This paper fills this gap,by first introducing the definition of NT and its causes,and reviewing over fifty representative approaches for overcoming NT,which fall into three categories:domain similarity estimation,safe transfer,and NT mitigation.Many areas,including computer vision,bioinformatics,natural language processing,recommender systems,and robotics,that use NT mitigation strategies to facilitate positive transfers,are also reviewed.Finally,we give guidelines on NT task construction and baseline algorithms,benchmark existing TL and NT mitigation approaches on three NT-specific datasets,and point out challenges and future research directions.To ensure reproducibility,our code is publicized at https://github.com/chamwen/NT-Benchmark.

Bilateral co-transfer for unsupervised domain adaptation

Labeled data scarcity of an interested domain is often a serious problem in machine learning.Leveraging the labeled data from other semantic-related yet co-variate shifted source domain to facilitate the interested domain is a consensus.In order to solve the domain shift between domains and reduce the learning ambiguity,unsupervised domain adaptation(UDA)greatly promotes the transferability of model parameters.However,the dilemma of over-fitting(negative transfer)and under-fitting(under-adaptation)is always an overlooked challenge and potential risk.In this paper,we rethink the shallow learning paradigm and this intractable over/under-fitting problem,and propose a safer UDA model,coined as Bilateral Co-Transfer(BCT),which is essentially beyond previous well-known unilateral transfer.With bilateral co-transfer between domains,the risk of over/under-fitting is therefore largely reduced.Technically,the proposed BCT is a symmetrical structure,with joint distribution discrepancy(JDD)modeled for domain alignment and category discrimination.Specifically,a symmetrical bilateral transfer(SBT)loss between source and target domains is proposed under the philosophy of mutual checks and balances.First,each target sample is represented by source samples with low-rankness constraint in a common subspace,such that the most informative and transferable source data can be used to alleviate negative transfer.Second,each source sample is symmetrically and sparsely represented by target samples,such that the most reliable target samples can be exploited to tackle underadaptation.Experiments on various benchmarks show that our BCT outperforms many previous outstanding work.

The Effect of the Native Language and Culture on EFL Writing of Chinese Students

According to an error analysis on Chinese college students' writing,the native language and culture have great influence onsecond-language writing.On the basis of error analysis,this paper put forward some countermeasures to improve Chinese students' Eng-lish writing skill.

Studies of negative pragmatic transfer in interlanguage pragmatics

Negative pragmatic transfer (NPT) is nothing but a difference of saying things between non native speakers and native speakers. It occupies an important position in interlanguage pragmatics whose mission is to scrutinize how non native speakers do things with words with L2. This paper reported that 4 NPT related aspects have been heavily documented in the current literature: 1) L1 negative pragmatic transfers at the speech act level; 2) the distinction between negative pragmalinguistic and sociopragmatic transfers; 3) conditions of negative pragmatic transfers; and 4) native speaker’s attitudes towards L1 negative pragmatic transfers. Consequently, issues for future studies are also raised.

The Negative Transfer from Xinyang Dialect on English Consonant Pronunciation

In the paper, the author presents the negative transfer influence of Xinyang dialect on learners' English consonant pronunciation by comparing their differences in the segmental level, and proposes some suggestions for English pronunciation teaching based on the theories of Contrastive Analysis, the Input Hypothesis.

A Study on Primary School English Learners' Consonant Epenthesis Errors

The study, with the reference book of PEP primary school, mainly focuses on the consonant epenthesis errors, analyzing the reasons behind the errors. Hereby it hopes to present some teaching solutions so as to help students to reduce the influence of negative transfer of mother tongue.

On L1 Negative Transfer in Second Language Acquisition Based on Chinese Students’Writing Composition

There is negative transfer of first language(L1)in second language acquisition,which generally introduces language errors,hindering learners’second language acquisition.Taking EFL learners’English composition as an example,this thesis analyzes the language errors caused by the negative transfer of L1.This thesis is conducted as the following four stages:The first part is an analysis of the relationship between language and thinking patterns,drawing forth the concept of language transfer;the second part is the literature review on second language acquisition and language transfer;the third part is the statistical analysis of the collected 110 samples;and the fourth part is the practical and innovative suggestion.This thesis aims to assist English acquirers in understanding the negative transfer of L1 and provide effective and practical measures in the future English teaching.

A Study on the Negative Phonetic Transfer of English Learners in Henan Province

Due to the influence of dialect,many English learners in China tend to speak English in the way they usually speak during second language acquisition,which results in negative transfer on English pronunciation.In order to reduce this phenomenon and improve the students’English pronunciation level,it is necessary for us to study the topic and put forward solutions.By using the contrastive analysis,this paper compares the pronunciation of English,Chinese and Henan dialect so as to find out the influence of Henan dialect on English learners at segmental level and suprasegmental level.The results show strong correlation between the pronunciation of English and Henan dialect for English learners.Therefore,this paper analyzes the mistakes and reasons that learners of Henan province are prone to make in the process of learning,and puts forward corresponding suggestions for teaching on students’negative transfer,so as to enhance the awareness of negative transfer for teachers and students.

The Influence of Negative Language Transfer

With the advance of the research of language learning, the field of SLA has expanded more expansively and native language transfer has become a popular issue in academia field. It exists in all language learning process, with different forms,which is mainly observed in three aspects including pronunciation, vocabulary and syntax. Furthermore, negative transfer is a major barrier influencing learners' second language improvement. This paper dwells on negative transfer. The comparison on language system and cultural background between native and target language should be attached great importance and analyzed.It is conducive to reduce the effect of native language and speed up the rapid learners' learning step.

An Analysis of Negative Transfer of Mother Tongue to Oral English and the Countermeasures

In the Chinese language market,English occupies a great position.The general public has gradually realized the impor⁃tance of mastering another language.However,due to the lack of language environment for learning English,mother tongue can eas⁃ily affect English learning production,which has two sides.Through the comparison between English and Chinese pronunciation,the negative transfer of native language in the acquisition process of English learners is analyzed,especially in oral English.Ac⁃cording to the existing problems,teaching suggestions are put forward for teachers’phonetic teaching and students’phonetic learn⁃ing to help learners reduce the negative effects of their mother tongue in the process of English acquisition.

On the Role of Distinctive Features of British English Sounds in the Cartoon Peppa Pig for Chinese Learners to Conquer Mother Tongue’s Negative Tra

With the process of globalization and integration,more and more people tend to be bilingual.Undoubtedly,mastering a second language is significant.This thesis aims to explore how to conquer the difficulties in learning British English sounds through analyzing a British cartoon Peppa Pig.Chapter one begins with the research background,significance and purpose of research.The thesis takes the cartoon Peppa Pig as the starting point to demonstrate the role of distinctive features for Chinese learners to conquer negative transfer.The empirical research could be found in Chapter two,because of requiring to know the concrete circumstance of Chinese learners.Through the collection and analysis of data,we can know these problems which Chinese learners exist in the process of learning British English.Based on the comparison of Chinese and British English,Chapter three clarifies the difficulties in learning British English.As for Chinese learners,mastering distinctive features can promote learners to master British English sounds better than stress and rhythm.Chapter four states the role of the distinctive features in helping conquer Chinese negative transfer in the cartoon Peppa Pig.The final chapter is the conclusion of this thesis and involves a new study for Chinese learners to learn British English sounds under the distinctive features.

The Influence of First Language Transfer on Second Language Acquisition

Language transfer plays a great role in second language acquisition process.As for the Chinese English learners,it is im⁃portant for them to know the positive and negative transfer of Chinese to English.Understanding the concepts of transfer and lan⁃guage transfer and the differences and similarities of mother language and target language may promote the acquisition of second language.According to the analysis of positive transfer and negative transfer,some learning strategies and pedagogical implications were made for second language learners.

Contrastive Analysis of Chinese and English Modes of Thinking and Its Effects on Discourse Organization in EFL Writing

Based on the contrastive analysis of Chinese and English modes of thinking, this essay analyzes from two aspects the effects of their discrepancies on the discourse organization of Chinese students ' English writing, then explains the fundamental reason for these effects and proposes that teachers should impart to students knowledge about the differences between Chinese and English mode of thinking so that Chinese students can be better helped to organize their compositions on the English discourse patterns and compose idiomatic English writings.

Study of Negative Effects of the Native Language Transfer on Second Language Acquisition

Language transfer has long been one of significant parts of the studies on the second language acquisition(SLA). The native language transfer will inevitably generate definite influences on SLA, negative or positive; it is less important to study the positive effects of the native language transfer. This paper is aimed to give a brief analysis on negative effects of the native language transfer in SLA with those English learners whose mother tongue is Chinese as examples. Meanwhile, in English teaching and learning, realization of negative transfers of the native language and utmost avoidance of them can improve the English learners' learning quality.

Chinese Students'Negative L1 Transfer in English Writing

Numerous researchers have studied the influence of L1 on the L2 composing process and consistently revealed that L2 writers may revert to their L1 when composing in L2(Cumming, 1989; Ellis, 2006;).The employment of L1 in L2 writing generally results in either facilitation(positive transfer) or impediment(negative transfer). The aim of this study is to investigate the specific errors and problems resulting from L1 interference within Chinese L2 writing and find possible solutions to minimize the negative effects of L1. In this respect, the focus of this paper is mainly on negative L1 transfer. Previous researches in this field have concentrated on how Chinese interfered L2 composing across different L2 proficiency levels or among different writing tasks. However, few studies have been carried out to analyze the details of how Chinese interfered with the students' English writing. This paper will conduct detailed analysis on the major negative transfer of L1 in terms of linguistic competence(from lexical, grammatical and syntactical level) and composing strategies(indirectness and organization style). The study will address and attempt to provide pedagogical suggestions for English writing instructors who particularly deal with Chinese L2 writers. The findings of the study has general applicability and will provide guidance for the teaching of L2 writing in other EFL/ESL settings.

Optimization of ohmic contact for InP-based transferred electronic devices

The effect of the annealing time and annealing temperature on Ni/Ge/Au electrode contacts deposited on the n-type InP contact layer has been studied using a circular transmission line model. The minimum specific contact resistance of 3.210 7 cm2was achieved on the low-doped n-type InP contact layer with a 40 s anneal at 425 ℃. In order to improve the ohmic contact and reduce the difficulty in the fabrication of the high doped InP epi-layer, the doping concentration in the InP contact layer was chosen to be 51018cm 3in the fabrication of transferred electronic devices. Excellent differential negative resistance properties were obtained by an electron beam evaporating the Ni/Ge/Au/Ge/Ni/Au composite electrode on an InP epi-layer with a 60 s anneal at 380 ℃.

Dual collaboration for decentralized multi-source domain adaptation

The goal of decentralized multi-source domain adaptation is to conduct unsupervised multi-source domain adaptation in a data decentralization scenario. The challenge of data decentralization is that the source domains and target domain lack cross-domain collaboration during training. On the unlabeled target domain, the target model needs to transfer supervision knowledge with the collaboration of source models, while the domain gap will lead to limited adaptation performance from source models. On the labeled source domain, the source model tends to overfit its domain data in the data decentralization scenario, which leads to the negative transfer problem. For these challenges, we propose dual collaboration for decentralized multi-source domain adaptation by training and aggregating the local source models and local target model in collaboration with each other. On the target domain, we train the local target model by distilling supervision knowledge and fully using the unlabeled target domain data to alleviate the domain shift problem with the collaboration of local source models. On the source domain, we regularize the local source models in collaboration with the local target model to overcome the negative transfer problem. This forms a dual collaboration between the decentralized source domains and target domain, which improves the domain adaptation performance under the data decentralization scenario. Extensive experiments indicate that our method outperforms the state-of-the-art methods by a large margin on standard multi-source domain adaptation datasets.