This paper focuses on bearing fault diagnosis with limited training data. A major challenge in fault diagnosis is the infeasibility of obtaining sufficient training samples for every fault type under all working conditions. Recently deep learning based fault diagnosis methods have achieved promising results. However, most of these methods require large amount of training data. In this study, we propose a deep neural network based few-shot learning approach for rolling bearing fault diagnosis with limited data. Our model is based on the siamese neural network, which learns by exploiting sample pairs of the same or different categories. Experimental results over the standard Case Western Reserve University (CWRU) bearing fault diagnosis benchmark dataset showed that our few-shot learning approach is more effective in fault diagnosis with limited data availability. When tested over different noise environments with minimal amount of training data, the performance of our few-shot learning model surpasses the one of the baseline with reasonable noise level. When evaluated over test sets with new fault types or new working conditions, few-shot models work better than the baseline trained with all fault types. All our models and datasets in this study are open sourced and can be downloaded from https://mekhub.cn/as/fault_diagnosis_with_few-shot_learning/. INDEX TERMS Deep learning, few-shot learning, bearing fault diagnosis, limited data.

Prognostics, such as remaining useful life (RUL) prediction, is a crucial task in condition-based maintenance. A major challenge in data-driven prognostics is the difficulty of obtaining a sufficient number of samples of failure progression. However, for traditional machine learning methods and deep neural networks, enough training data is a prerequisite to train good prediction models. In this work, we proposed a transfer learning algorithm based on Bi-directional Long Short-Term Memory (BLSTM) recurrent neural networks for RUL estimation, in which the models can be first trained on different but related datasets and then fine-tuned by the target dataset. Extensive experimental results show that transfer learning can in general improve the prediction models on the dataset with a small number of samples. There is one exception that when transferring from multi-type operating conditions to single operating conditions, transfer learning led to a worse result.

No abstract

Interactions between human leukocyte antigens (HLAs) and peptides play a critical role in the human immune system. Accurate computational prediction of HLA-binding peptides can be used for peptide drug discovery. Currently, the best prediction algorithms are neural network-based pan-specific models, which take advantage of the large amount of data across HLA alleles. However, current pan-specific models are all based on the pseudo sequence encoding for modeling the binding context, which is based on 34 positions identified from the HLA protein-peptide bound structures in early works. In this work, we proposed a novel deep convolutional neural network model (DCNN) for HLA-peptide binding prediction, in which the encoding of the HLA sequence and the binding context are both learned by the network itself without requiring the HLA-peptide bound structure information. Our DCNN model is also characterized by its binding context extraction layer and dual outputs with both binding affinity output and binding probability outputs. Evaluation on public benchmark datasets shows that our DeepSeqPan model without HLA structural information in training achieves state-of-the-art performance on a large number of HLA alleles with good generalization capability. Since our model only needs raw sequences from the HLA-peptide binding pairs, it can be applied to binding predictions of HLAs without structure information and can also be applied to other protein binding problems such as protein-DNA and protein-RNA bindings. The implementation code and trained models are freely available at https://github.com/pcpLiu/DeepSeqPan.