Predicting Protein Functions of Bacteria Genomes via Multi-instance Multi-Label Active Learning

Wu, Jiansheng; Zhu, Wei-Ping; Jiang, Ye; Sun, Guwei; Gao, Yusheng

doi:10.1109/icam.2018.8596617

Cited by 5 publications

(4 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In their methods, not only one selected label is queried, but also the key instance which is most relevant to queried label is asked. In [36], the authors extend work in [35] with a modification in bag label prediction achieving from instance-level predictions to predicting protein functions of bacteria genomes. All these approaches select a bag to label all classes or a bag and then select a class to label.…”

Section: Related Workmentioning

confidence: 99%

“…Finally, by substituting the bound on f 0 (0) in ( 43) in (36), we obtain the following bound on the l 2 -norm of the optimal parameter vector:…”

Section: A Bounding the Optimal Parameter Vectormentioning

confidence: 99%

“…(ii) A bag is selected first and then a single class for that bag is selected to be labeled at each query [35], [36]. In [35], [36], after selecting the bag and the class to be labeled, the oracle decides whether they are relevant. If the oracle returns a negative label the query process is complete and all instances are assumed negative for the selected class.…”

Section: Introductionmentioning

confidence: 99%

“…This approach may lead to redundant labeling of classes, which do not help to increase the performance of the model. (ii) A bag is selected first and then a single class for that bag is selected to be labeled at each query [35], [36]. In [35], [36], after selecting the bag and the class to be labeled, the oracle decides whether they are relevant.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Incomplete Label Multiple Instance Multiple Label Learning

Nguyen

Raich

2022

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

To alleviate labeling complexity, in multi-instance multi-label learning, each sample/bag consists of multiple instances and is associated with a set of bag-level labels leaving instances therein unlabeled. This setting is more convenient and natural for representing complicated objects with multiple semantic meanings. Compared to single-instance labeling, this approach allows for labeling larger datasets at an equivalent labeling cost. However, for sufficiently large datasets, labeling all bags may become prohibitively costly. Active learning (AL) uses an iterative labeling and retraining approach to provide reasonable classification performance using a small number of labeled samples. To our knowledge, only two approaches have been previously proposed for AL in the MIML setting. These approaches either require labeling all classes in a selected bag or involve partial instance-level labeling. To further reduce labeling costs, we propose a novel bag-class pair-based approach for AL in the MIML setting. Due to the partial availability of bag-level labels, we focus on AL in the incomplete-label MIML setting. For the query process, we adapt AL criteria to the novel bag-class pair selection strategy. Additionally, we introduce an online approach for learning a discriminative graphical model based classifier. Numerical experiments on benchmark datasets demonstrate the effectiveness of the proposed approach.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Finally, by substituting the bound on f 0 (0) in ( 43) in (36), we obtain the following bound on the l 2 -norm of the optimal parameter vector:…”

Section: A Bounding the Optimal Parameter Vectormentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Incomplete Label Multiple Instance Multiple Label Learning

Nguyen

Raich

2022

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

show abstract

Machine learning techniques for protein function prediction

2019

View full text Add to dashboard Cite

Proteins play important roles in living organisms, and their function is directly linked with their structure. Due to the growing gap between the number of proteins being discovered and their functional characterization (in particular as a result of experimental limitations), reliable prediction of protein function through computational means has become crucial. This paper reviews the machine learning techniques used in the literature, following their evolution from simple algorithms such as logistic regression to more advanced methods like support vector machines and modern deep neural networks. Hyperparameter optimization methods adopted to boost prediction performance are presented. In parallel, the metamorphosis in the features used by these algorithms from classical physicochemical properties and amino acid composition, up to text‐derived features from biomedical literature and learned feature representations using autoencoders, together with feature selection and dimensionality reduction techniques, are also reviewed. The success stories in the application of these techniques to both general and specific protein function prediction are discussed.

show abstract