Transfer learning using computational intelligence

Question

1. What have the authors contributed in "Transfer learning using computational intelligence: a survey" ?

2. What are the two phases of learning a Bayesian network from data?

3. What is the main purpose of the self-training method?

4. What are the main reasons why researchers take fuzzy systems into account for transfer learning more and more?

Accepted Answer

This paper systematically examines computational intelligence-based transfer learning techniques and clusters related technique developments into four main categories: a ) neural network-based transfer learning ; b ) Bayes-based transfer learning ; c ) fuzzy transfer learning, and d ) applications of computational intelligence-based transfer learning.. By providing state-of-the-art knowledge, this survey will directly support researchers and practice-based professionals to understand the developments in computational intelligence-based transfer learning research and applications.

Accepted Answer

To learn a Bayesian network from data, one needs to consider two important phases: structure learning and parameter learning, respectively.

Accepted Answer

Self-training methods have been applied to domain adaptation on Natural Language Processing (NLP) tasks including parsing [18-21]; part-of-speech tagging [22]; conversation summarization [23]; entity recognition [22, 24, 25]; sentiment classification [26]; spam detection [22]; cross-language document classification [27, 28]; and speech act classification [29].

Accepted Answer

Since many real world applications have noisy and uncertainty in data, researchers take fuzzy systems into account for transfer learning more and more.

Accepted Answer

Transfer learning approaches with the support of computational intelligence methods such as neural network, Bayesian network, and fuzzy logic have been applied in real-world applications.

Accepted Answer

The fundamental motivation for transfer learning in the field of machine learning focuses on the need for lifelong machine learning methods that retain and reuse previously learned knowledge.

Accepted Answer

The main limitation of such multi-task network structure learning algorithms lies in the assumption that all pairs of tasks are equally related, which violates the truth that different pairs of tasks can differ in their degree of relatedness.

Accepted Answer

Research on transfer learning has been undertaken since 1995 under a variety of names: learning to learn; life-long learning; knowledge transfer; meta learning; inductive transfer; knowledge consolidation; context sensitive learning and multi-task learning [9].

Accepted Answer

Using fuzzy techniques in similarity measurement and label production, the authors revealed the advantage of fuzzy logic in knowledge transfer where the target domain lacks critical information and involves uncertainty and vagueness.

Accepted Answer

By applying salient feature detection and tracking in videos to simulate fixations and smooth pursuit in human vision, Zou et al. [102] successfully implemented an unsupervised learning algorithm in a self-taught learning setting.

Accepted Answer

Of the computational intelligence methods, neural network has been extensively used for transfer learning, mainly in computer vision and image processing domain.

Accepted Answer

Transfer learning has emerged in the computer science literature as a means of transferring knowledge from a source domain to a target domain.

Accepted Answer

The main reason why neural network has been widely used in transfer learning is that it doesn’t have I.I.D. assumption on data while almost all stochastic techniques have.

Accepted Answer

The two primary elements within fuzzy logic, the linguistic variable and the fuzzy if-then rule, are able to mimic the human ability to capture imprecision and uncertainty within knowledge transfer.

Accepted Answer

From the summary of transfer learning, it is concluded that transfer learning with the use computational intelligence, as an emerging research topic, starts playing an important role in almost all kinds of application.

Accepted Answer

Chopra et al. [51] argued that in the representation learning camp for images, existing deep learning approaches could not encode the distributional shift between the source and target domains.

Accepted Answer

As a result, once the feature space or the feature distribution of the test data changes, the prediction models cannot be used and must be rebuilt and retrained from scratch using newly-collected training data, which is very expensive and sometimes not practically possible.

Accepted Answer

Authors [50] also considered a problem of classifying images of English lowercase a-to-z by reusing fine-tuned features of English handwritten digits 0-to-9.

Accepted Answer

Koçer and Arslan [97] introduced the use of genetic algorithm and transfer learning by extending a previously constructed algorithm.

Accepted Answer

since learning-based models need adequate labeled data for training, it is nearly impossible to establish a learning-based model for a target domain which has very few labeled data available for supervised learning.

Transfer learning using computational intelligence

Chat with Paper

AI Agents for this Paper

Most frequently asked questions

1. What have the authors contributed in "Transfer learning using computational intelligence: a survey" ?

2. What are the two phases of learning a Bayesian network from data?

3. What is the main purpose of the self-training method?

4. What are the main reasons why researchers take fuzzy systems into account for transfer learning more and more?

5. What are the main applications of transfer learning?

6. What is the fundamental motivation for transfer learning in the field of machine learning?

7. What is the main limitation of multi-task network structure learning algorithms?

8. What is the main purpose of transfer learning?

9. What is the advantage of fuzzy logic in knowledge transfer?

10. How did Zou et al. achieve this?

11. What are the main reasons why neural network has been widely used for transfer learning?

12. What is the purpose of transfer learning?

13. What is the main reason why neural network has been widely used in transfer learning?

14. What is the main difference between fuzzy logic and classical transductive transfer learning?

15. What is the main reason why the research paper is a summary of transfer learning?

16. What is the problem in the representation learning camp for images?

17. What is the cost of retraining a prediction model?

18. What is the problem of reusing English lowercase digits?

19. How did they extend a previously constructed algorithm?

20. What is the main reason for the lack of labeled training data?

Figures

Citations

A Comprehensive Survey on Transfer Learning

Automatic detection of coronavirus disease (COVID-19) using X-ray images and deep convolutional neural networks.

Automatic Detection of Coronavirus Disease (COVID-19) Using X-ray Images and Deep Convolutional Neural Networks

A New Deep Transfer Learning Based on Sparse Auto-Encoder for Fault Diagnosis

A Survey of Unsupervised Deep Domain Adaptation

References

Fuzzy sets

Sample Selection Bias as a Specification Error

A Survey on Transfer Learning

Receptive fields, binocular interaction and functional architecture in the cat's visual cortex

Detecting change in vague interpretations of landscapes

Related Papers (5)

A Survey on Transfer Learning

Deep learning

Deep Residual Learning for Image Recognition

Domain-adversarial training of neural networks

ImageNet Classification with Deep Convolutional Neural Networks