Search

Medical images are valuable for clinical diagnosis and decision making. Image modality is an important primary step, as it is capable of aiding clinicians to access required medical image in retrieval systems. Traditional methods of modality classification are dependent on the choice of hand-crafted features and demand a clear awareness of prior domain knowledge. The feature learning approach may detect efficiently visual characteristics of different modalities, but it is limited to the number of training datasets. To overcome the absence of labeled data, on the one hand, we take deep convolutional neural networks (VGGNet, ResNet) with different depths pre-trained on ImageNet, fix most of the earlier layers to reserve generic features of natural images, and only train their higher-level portion on ImageCLEF to learn domain-specific features of medical figures. Then, we train from scratch deep CNNs with only six weight layers to capture more domain-specific features. On the other hand, we employ two data augmentation methods to help CNNs to give the full scope to their potential characterizing image modality features. The final prediction is given by our voting system based on the outputs of three CNNs. After evaluating our proposed model on the subfigure classification task in ImageCLEF2015 and ImageCLEF2016, we obtain new, state-of-the-art results—76.87% in ImageCLEF2015 and 87.37% in ImageCLEF2016—which imply that CNNs, based on our proposed transfer learning methods and data augmentation skills, can identify more efficiently modalities of medical images.

The literature [-5]contains several reports evaluating the abilities of deep neural networks in text transfer learning. To our knowledge, however, there have been few efforts to fully realize the potential of deep neural networks in cross-domain product review sentiment classification. In this paper, we propose a two-layer convolutional neural network (CNN) for cross-domain product review sentiment classification (LM-CNN-LB). Transfer learning research into product review sentiment classification based on deep neural networks has been limited by the lack of a large-scale corpus; we sought to remedy this problem using a large-scale auxiliary cross-domain dataset collected from Amazon product reviews. Our proposed framework exhibits the dramatic transferability of deep neural networks for cross-domain product review sentiment classification and achieves state-of-the-art performance. The framework also outperforms complex engineered features used with a non-deep neural network method. The experiments demonstrate that introducing large-scale data from similar domains is an effective way to resolve the lack of training data. The LM-CNN-LB trained on the multi-source related domain dataset outperformed the one trained on a single similar domain.

Learners in a massive open online course often express feelings, exchange ideas and seek help by posting questions in discussion forums. Due to the very high learner-to-instructor ratios, it is unrealistic to expect instructors to adequately track the forums, find all of the issues that need resolution and understand their urgency and sentiment. In this paper, considering the biases among different courses, we propose a transfer learning framework based on a convolutional neural network and a long short-term memory model, called ConvL, to automatically identify whether a post expresses confusion, determine the urgency and classify the polarity of the sentiment. First, we learn the feature representation for each word by considering the local contextual feature via the convolution operation. Second, we learn the post representation from the features extracted through the convolution operation via the LSTM model, which considers the long-term temporal semantic relationships of features. Third, we investigate the possibility of transferring parameters from a model trained on one course to another course and the subsequent fine-tuning. Experiments on three real-world MOOC courses confirm the effectiveness of our framework. This work suggests that our model can potentially significantly increase the effectiveness of monitoring MOOC forums in real time.

Compound figure detection on figures and associated captions is the first step to making medical figures from biomedical literature available for further analysis. The performance of traditional methods is limited to the choice of hand-engineering features and prior domain knowledge. We train multiple convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and gated recurrent unit (GRU) networks on top of pre-trained word vectors to learn textual features from captions and employ deep CNNs to learn visual features from figures. We then identify compound figures by combining textual and visual prediction. Our proposed architecture obtains remarkable performance in three run types—textual, visual and mixed—and achieves better performance in ImageCLEF2015 and ImageCLEF2016.

We present a new music retrieval system based on query by semantic description (QBSD) system, by which a novel song can be used as query and transformed into semantic vector by a convolutional neural network. This method based on Supervised Multi-class labeling (SML), which a song can be annotated by some semantically meaningful tags and retrieved relevant song in semantically annotated database. CAL500 data set is used in experiment, we can learn a deep learning model for each tag in semantic space. To improve the annotation effect, loss function adjustment algorithm and SMOTE algorithm are employed. The experiment results show that this model can get songs with high semantically similarity, and provide a more nature way to music retrieval.

Text mining tasks based on machine learning require inputs to be represented as fixed-length vectors, and effective vectors of words, phrases, sentences and even documents may greatly improve the performance of these tasks. Recently, distributed word representations based on neural networks have been demonstrated powerful in many tasks by encoding abundant semantic and linguistic information. However, it remains a great challenge for document representations because of the complex semantic structures in different documents. To meet the challenge, we propose two novel tripartite graphical models for document representations by incorporating word representations into the Replicated Softmax model, and we name the models as Tripartite-Replicated Softmax model (TRPS) and directed Tripartite-Replicated Softmax model (d-TRPS), respectively. We also introduce some optimization strategies for training the proposed models to learn better document representations. The proposed models can capture linear relationships among words and latent semantic information within documents simultaneously, thus learning both linear and nonlinear document representations. We examine the learned document representations in a document classification task and a document retrieval task. Experimental results show that the learned representations by our models outperform the state-of-the-art models in improving the performance of these two tasks.