Search

Question answering (QA)-based re-ranking methods for cross-modal retrieval have been recently proposed to further narrow down similar candidate images. The conventional QA-based re-ranking methods provide questions to users by analyzing candidate images, and the initial retrieval results are re-ranked based on the user's feedback. Contrary to these developments, only focusing on performance improvement makes it difficult to efficiently elicit the user's retrieval intention. To realize more useful QA-based re-ranking, considering the user interaction for eliciting the user's retrieval intention is required. In this paper, we propose a QA-based re-ranking method with considering two important factors for eliciting the user's retrieval intention: query-image relevance and recallability. Considering the query-image relevance enables to only focus on the candidate images related to the provided query text, while, focusing on the recallability enables users to easily answer the provided question. With these procedures, our method can efficiently and effectively elicit the user's retrieval intention. Experimental results using Microsoft Common Objects in Context and computationally constructed dataset including similar candidate images show that our method can improve the performance of the cross-modal retrieval methods and the QA-based re-ranking methods.

This study presents a method for distress image classification in road infrastructures introducing self-supervised learning. Self-supervised learning is an unsupervised learning method that does not require class labels. This learning method can reduce annotation efforts and allow the application of machine learning to a large number of unlabeled images. We propose a novel distress image classification method using contrastive learning, which is a type of self-supervised learning. Contrastive learning provides image domain-specific representation, constraining such that similar images are embedded nearby in the latent space. We augment the single input distress image into multiple images by image transformations and construct the latent space, in which the augmented images are embedded close to each other. This provides a domain-specific representation of the damage in road infrastructure using a large number of unlabeled distress images. Finally, the representation obtained by contrastive learning is used to improve the distress image classification performance. The obtained contrastive learning model parameters are used for the distress image classification model. We realize the successful distress image representation by utilizing unlabeled distress images, which have been difficult to use in the past. In the experiments, we use the distress images obtained from the real world to verify the effectiveness of the proposed method for various distress types and confirm the performance improvement.

A cross-modal image retrieval that explicitly considers semantic relationships between images and texts is proposed. Most conventional cross-modal image retrieval methods retrieve the target images by directly measuring the similarities between the candidate images and query texts in a common semantic embedding space. However, such methods tend to focus on a one-to-one correspondence between a predefined image-text pair during the training phase, and other semantically similar images and texts are ignored. By considering the many-to-many correspondences between semantically similar images and texts, a common embedding space is constructed to assure semantic relationships, which allows users to accurately find more images that are related to the input query texts. Thus, in this paper, we propose a cross-modal image retrieval method that considers semantic relationships between images and texts. The proposed method calculates the similarities between texts as semantic similarities to acquire the relationships. Then, we introduce a loss function that explicitly constructs the many-to-many correspondences between semantically similar images and texts from their semantic relationships. We also propose an evaluation metric to assess whether each method can construct an embedding space considering the semantic relationships. Experimental results demonstrate that the proposed method outperforms conventional methods in terms of this newly proposed metric.

Auxiliary clinical diagnosis has been researched to solve unevenly and insufficiently distributed clinical resources. However, auxiliary diagnosis is still dominated by human physicians, and how to make intelligent systems more involved in the diagnosis process is gradually becoming a concern. An interactive automated clinical diagnosis with a question-answering system and a question generation system can capture a patient's conditions from multiple perspectives with less physician involvement by asking different questions to drive and guide the diagnosis. This clinical diagnosis process requires diverse information to evaluate a patient from different perspectives to obtain an accurate diagnosis. Recently proposed medical question generation systems have not considered diversity. Thus, we propose a diversity learning-based visual question generation model using a multi-latent space to generate informative question sets from medical images. The proposed method generates various questions by embedding visual and language information in different latent spaces, whose diversity is trained by our newly proposed loss. We have also added control over the categories of generated questions, making the generated questions directional. Furthermore, we use a new metric named similarity to accurately evaluate the proposed model's performance. The experimental results on the Slake and VQA-RAD datasets demonstrate that the proposed method can generate questions with diverse information. Our model works with an answering model for interactive automated clinical diagnosis and generates datasets to replace the process of annotation that incurs huge labor costs.

This study presents a method for distress image classification in road infrastructures introducing self-supervised learning. Self-supervised learning is an unsupervised learning method that does not require class labels. This learning method can reduce annotation efforts and allow the application of machine learning to a large number of unlabeled images. We propose a novel distress image classification method using contrastive learning, which is a type of self-supervised learning. Contrastive learning provides image domain-specific representation, constraining such that similar images are embedded nearby in the latent space. We augment the single input distress image into multiple images by image transformations and construct the latent space, in which the augmented images are embedded close to each other. This provides a domain-specific representation of the damage in road infrastructure using a large number of unlabeled distress images. Finally, the representation obtained by contrastive learning is used to improve the distress image classification performance. The obtained contrastive learning model parameters are used for the distress image classification model. We realize the successful distress image representation by utilizing unlabeled distress images, which have been difficult to use in the past. In the experiments, we use the distress images obtained from the real world to verify the effectiveness of the proposed method for various distress types and confirm the performance improvement.

A cross-modal image retrieval that explicitly considers semantic relationships between images and texts is proposed. Most conventional cross-modal image retrieval methods retrieve the target images by directly measuring the similarities between the candidate images and query texts in a common semantic embedding space. However, such methods tend to focus on a one-to-one correspondence between a predefined image-text pair during the training phase, and other semantically similar images and texts are ignored. By considering the many-to-many correspondences between semantically similar images and texts, a common embedding space is constructed to assure semantic relationships, which allows users to accurately find more images that are related to the input query texts. Thus, in this paper, we propose a cross-modal image retrieval method that considers semantic relationships between images and texts. The proposed method calculates the similarities between texts as semantic similarities to acquire the relationships. Then, we introduce a loss function that explicitly constructs the many-to-many correspondences between semantically similar images and texts from their semantic relationships. We also propose an evaluation metric to assess whether each method can construct an embedding space considering the semantic relationships. Experimental results demonstrate that the proposed method outperforms conventional methods in terms of this newly proposed metric.

Auxiliary clinical diagnosis has been researched to solve unevenly and insufficiently distributed clinical resources. However, auxiliary diagnosis is still dominated by human physicians, and how to make intelligent systems more involved in the diagnosis process is gradually becoming a concern. An interactive automated clinical diagnosis with a question-answering system and a question generation system can capture a patient's conditions from multiple perspectives with less physician involvement by asking different questions to drive and guide the diagnosis. This clinical diagnosis process requires diverse information to evaluate a patient from different perspectives to obtain an accurate diagnosis. Recently proposed medical question generation systems have not considered diversity. Thus, we propose a diversity learning-based visual question generation model using a multi-latent space to generate informative question sets from medical images. The proposed method generates various questions by embedding visual and language information in different latent spaces, whose diversity is trained by our newly proposed loss. We have also added control over the categories of generated questions, making the generated questions directional. Furthermore, we use a new metric named similarity to accurately evaluate the proposed model's performance. The experimental results on the Slake and VQA-RAD datasets demonstrate that the proposed method can generate questions with diverse information. Our model works with an answering model for interactive automated clinical diagnosis and generates datasets to replace the process of annotation that incurs huge labor costs.

Question answering (QA)-based re-ranking methods for cross-modal retrieval have been recently proposed to further narrow down similar candidate images. The conventional QA-based re-ranking methods provide questions to users by analyzing candidate images, and the initial retrieval results are re-ranked based on the user's feedback. Contrary to these developments, only focusing on performance improvement makes it difficult to efficiently elicit the user's retrieval intention. To realize more useful QA-based re-ranking, considering the user interaction for eliciting the user's retrieval intention is required. In this paper, we propose a QA-based re-ranking method with considering two important factors for eliciting the user's retrieval intention: query-image relevance and recallability. Considering the query-image relevance enables to only focus on the candidate images related to the provided query text, while, focusing on the recallability enables users to easily answer the provided question. With these procedures, our method can efficiently and effectively elicit the user's retrieval intention. Experimental results using Microsoft Common Objects in Context and computationally constructed dataset including similar candidate images show that our method can improve the performance of the cross-modal retrieval methods and the QA-based re-ranking methods.

Dataset complexity assessment aims to predict classification performance on a dataset with complexity calculation before training a classifier, which can also be used for classifier selection and dataset reduction. The training process of deep convolutional neural networks (DCNNs) is iterative and time-consuming because of hyperparameter uncertainty and the domain shift introduced by different datasets. Hence, it is meaningful to predict classification performance by assessing the complexity of datasets effectively before training DCNN models. This paper proposes a novel method called cumulative maximum scaled Area Under Laplacian Spectrum (cmsAULS), which can achieve state-of-the-art complexity assessment performance on six datasets.

This paper presents deterioration level estimation based on convolutional neural networks using a confidence-aware attention mechanism for infrastructure inspection. Spatial attention mechanisms try to highlight the important regions in feature maps for estimation by using an attention map. The attention mechanism using an effective attention map can improve feature maps. However, the conventional attention mechanisms have a problem as they fail to highlight important regions for estimation when an ineffective attention map is mistakenly used. To solve the above problem, this paper introduces the confidence-aware attention mechanism that reduces the effect of ineffective attention maps by considering the confidence corresponding to the attention map. The confidence is calculated from the entropy of the estimated class probabilities when generating the attention map. Because the proposed method can effectively utilize the attention map by considering the confidence, it can focus more on the important regions in the final estimation. This is the most significant contribution of this paper. The experimental results using images from actual infrastructure inspections confirm the performance improvement of the proposed method in estimating the deterioration level.

In this paper, we present refining graph representation for cross-domain recommendation (CDR) based on edge pruning considering feature distribution in a latent space. Conventional graph-based CDR methods have utilized all ratings and purchase histories of user's products. However, some items purchased by users are not related to the domain for recommendation, and this information becomes noise when making CDR. So, the proposed method introduces edge pruning into the latest graph-based CDR method to refine graph representation. To compare the item embedding features calculated in different domains, we construct a latent space and perform edge pruning through their correlations. Additionally, we introduce a state-of-the-art graph neural network into the graph construction of the proposed method that considers the interactions between users and items thereby obtaining effective embedding features in a domain. This makes it possible to consider domain-specific user preferences and estimate embedding features with high-expressive power. Furthermore, to compare the embedding features of items in the two domains, we construct their latent spaces and project them. Edge pruning is performed using the correlation of items between the two domains on the latent space. We obtain cross domain specific graph representation through edge pruning, which improves the performance by considering the relationship between both items across domains. To the best of our knowledge, no study in the CDR field focuses on eliminating unnecessary node information. We have demonstrated the effectiveness of the proposed method by comparing several graph-based state-of-the-art methods.

In this study, a novel prediction method for predicting important scenes in baseball videos using a time-lag aware latent variable model (Tl-LVM) is proposed. Tl-LVM adopts a multimodal variational autoencoder using tweets and videos as the latent variable model. It calculates the latent features from these tweets and videos and predicts important scenes using these latent features. Since time lags exist between posted tweets and events, Tl-LVM introduces the loss considering time lags by correlating the feature into the loss function of the multimodal variational autoencoder. Furthermore, Tl-LVM can train the encoder, decoder, and important scene predictor, simultaneously, using this loss function. This is the novelty of Tl-LVM, and this work is the first end-to-end prediction model of important scenes that considers time lags to the best of our knowledge. It is the contribution of Tl-LVM to realize high-quality prediction using latent features that consider time lags between tweets and multiple corresponding previous events. Experimental results using actual tweets and baseball videos show the effectiveness of Tl-LVM.

In this study, we propose a novel music playlist generation method based on a knowledge graph and reinforcement learning. The development of music streaming platforms has transformed the social dynamics of music consumption and paved a new way of accessing and listening to music. The playlist generation is one of the most important multimedia techniques, which aims to recommend music tracks by sensing the vast amount of musical data and the users' listening histories from music streaming services. Conventional playlist generation methods have difficulty capturing the target users' long-term preferences. To overcome the difficulty, we use a reinforcement learning algorithm that can consider the target users' long-term preferences. Furthermore, we introduce the following two new ideas: using the informative knowledge graph data to promote efficient optimization through reinforcement learning, and setting the flexible reward function that target users can design the parameters of itself to guide target users to new types of music tracks. We confirm the effectiveness of the proposed method by verifying the prediction performance based on listening history and the guidance performance to music tracks that can satisfy the target user's unique preference.

Regularization has become an important method in adversarial defense. However, the existing regularization-based defense methods do not discuss which features in convolutional neural networks (CNN) are more suitable for regularization. Thus, in this paper, we propose a multi-stage feature fusion network with a feature regularization operation, which is called Enhanced Multi-Stage Feature Fusion Network (EMSF(2)Net). EMSF(2)Net mainly combines three parts: multi-stage feature enhancement (MSFE), multi-stage feature fusion (MSF2), and regularization. Specifically, MSFE aims to obtain enhanced and expressive features in each stage by multiplying the features of each channel; MSF2 aims to fuse the enhanced features of different stages to further enrich the information of the feature, and the regularization part can regularize the fused and original features during the training process. EMSF(2)Net has proved that if the regularization term of the enhanced multi-stage feature is added, the adversarial robustness of CNN will be significantly improved. The experimental results on extensive white-box attacks on the CIFAR-10 dataset illustrate the robustness and effectiveness of the proposed method.

Brain decoding is a process of decoding human cognitive contents from brain activities. However, improving the accuracy of brain decoding remains difficult due to the unique characteristics of the brain, such as the small sample size and high dimensionality of brain activities. Therefore, this paper proposes a method that effectively uses multi-subject brain activities to improve brain decoding accuracy. Specifically, we distinguish between the shared information common to multi-subject brain activities and the individual information based on each subject's brain activities, and both types of information are used to decode human visual cognition. Both types of information are extracted as features belonging to a latent space using a probabilistic generative model. In the experiment, an publicly available dataset and five subjects were used, and the estimation accuracy was validated on the basis of a confidence score ranging from 0 to 1, and a large value indicates superiority. The proposed method achieved a confidence score of 0.867 for the best subject and an average of 0.813 for the five subjects, which was the best compared to other methods. The experimental results show that the proposed method can accurately decode visual cognition compared with other existing methods in which the shared information is not distinguished from the individual information.

Distresses, such as cracks, directly reflect the structural integrity of subway tunnels. Therefore, the detection of subway tunnel distress is an essential task in tunnel structure maintenance. This paper presents the performance improvement of deep learning-based distress detection to support the maintenance of subway tunnels through a new data augmentation method, selective image cropping and patching (SICAP). Specifically, we generate effective data for training the distress detection model by focusing on the distressed regions via SICAP. After the data augmentation, we train a distress detection model using the expanded training data. The new image generated based on SICAP does not change the pixel values of the original image. Thus, there is little loss of information, and the generated images are effective in constructing a robust model for various subway tunnel lines. We conducted experiments with some comparative methods. The experimental results show that the detection performance can be improved by our data augmentation.

Dataset complexity assessment aims to predict classification performance on a dataset with complexity calculation before training a classifier, which can also be used for classifier selection and dataset reduction. The training process of deep convolutional neural networks (DCNNs) is iterative and time-consuming because of hyperparameter uncertainty and the domain shift introduced by different datasets. Hence, it is meaningful to predict classification performance by assessing the complexity of datasets effectively before training DCNN models. This paper proposes a novel method called cumulative maximum scaled Area Under Laplacian Spectrum (cmsAULS), which can achieve state-of-the-art complexity assessment performance on six datasets.

A human-centric emotion estimation method based on correlation maximization with consideration of changes with time in visual attention and brain activity when viewing images is proposed in this paper. Owing to the recent developments of many kinds of biological sensors, many researchers have focused on multimodal emotion estimation using both eye gaze data and brain activity data for improving the quality of emotion estimation. In this paper, a novel method that focuses on the following two points is introduced. First, in order to reduce the burden on users, we obtain brain activity data from users only in the training phase by using a projection matrix calculated by canonical correlation analysis (CCA) between gaze-based visual features and brain activity-based features. Next, for considering the changes with time in both visual attention and brain activity, we obtain novel features based on CCA-based projection in each time unit. In order to include these two points, the proposed method analyzes a fourth-order gaze and image tensor for which modes are pixel location, color channel and the changes with time in visual attention. Moreover, in each time unit, the proposed method performs CCA between gaze-based visual features and brain activity-based features to realize human-centric emotion estimation with a high level of accuracy. Experimental results show that accurate human emotion estimation is achieved by using our new human-centric image representation.

Distress image retrieval for infrastructure maintenance via self-trained deep metric learning using experts' knowledge is proposed in this paper. Since engineers take multiple images of a single distress part for inspection of road structures, it is necessary to construct a similar distress image retrieval method considering the input of multiple images to support determination of the level of deterioration. Thus, the construction of an image retrieval method while selecting an effective input from multiple images is described in this paper. The proposed method performs deep metric learning by using a small number of effective images labeled by experts' knowledge with information about their effectiveness and a large number of unlabeled images via a self-training approach. Specifically, an end-to-end learning approach that performs retraining of the model by assigning pseudo-labels to these unlabeled images according to the output confidence of the model is achieved. Thus, the proposed method can select an effective image from multiple images that are input at the retrieval as a query image. This is the main contribution of this paper. As a result, the proposed method realizes highly accurate retrieval of similar distress images considering the actual situation of inspection in which multiple images of a distress part are input.

A new style transfer-based image manipulation framework combining generative networks and style transfer networks is presented in this paper. Unlike conventional style transfer tasks, we tackle a new task, text-guided image manipulation. We realize style transfer-based image manipulation that does not require any reference style images and generate a style image from the user's input sentence. In our method, since an initial reference input sentence for a content image can automatically be given by an image-to-text model, the user only needs to update the reference sentence. This scheme can help users when they do not have any images representing the desired style. Although this text-guided image manipulation is a new challenging task, quantitative and qualitative comparisons showed the superiority of our method.

A method for prediction of properties of rubber materials utilizing electron microscope images of internal structures taken under multiple conditions is presented in this paper. Electron microscope images of rubber materials are taken under several conditions, and effective conditions for the prediction of properties are different for each rubber material. Novel approaches for the selection and integration of reliable prediction results are used in the proposed method. The proposed method enables selection of reliable results based on prediction intervals that can be derived by the predictors that are each constructed from electron microscope images taken under each condition. By monitoring the relationship between prediction results and prediction intervals derived from the corresponding predictors, it can be determined whether the target prediction results are reliable. Furthermore, the proposed method integrates the selected reliable results based on Dempster-Shafer (DS) evidence theory, and this integration result is regarded as a final prediction result. The DS evidence theory enables integration of multiple prediction results, even if the results are obtained from different imaging conditions. This means that integration can even be realized if electron microscope images of each material are taken under different conditions and even if these conditions are different for target materials. This nonconventional approach is suitable for our application, i.e., property prediction. Experiments on rubber material data showed that the evaluation index mean absolute percent error (MAPE) was under 10% by the proposed method. The performance of the proposed method outperformed conventional comparative property estimation methods. Consequently, the proposed method can realize accurate prediction of the properties with consideration of the characteristic of electron microscope images described above.

A new method for the detection of important scenes in baseball videos via a time-lag-aware multimodal variational autoencoder (Tl-MVAE) is presented in this paper. Tl-MVAE estimates latent features calculated from tweet, video, and audio features extracted from tweets and videos. Then, important scenes are detected by estimating the probability of the scene being important from estimated latent features. It should be noted that there exist time-lags between tweets posted by users and videos. To consider the time-lags between tweet features and other features calculated from corresponding multiple previous events, the feature transformation based on feature correlation considering such time-lags is newly introduced to the encoder in MVAE in the proposed method. This is the biggest contribution of the Tl-MVAE. Experimental results obtained from actual baseball videos and their corresponding tweets show the effectiveness of the proposed method.

This study presents a novel feature integration method for interest level estimation using a semi-supervised multimodal Gaussian process latent variable model with pseudo-labels (semi-MGPPL). Semi-MGPPL is an extended version of the multimodal Gaussian process latent variable model (mGPLVM). It integrates features calculated from multiple modalities to predict the users' interest levels in content. It is known that reflecting known interest levels of known users in the latent space effectively improves the accuracy of interest level estimation. However, previous methods have difficulty reflecting the interest levels when the number of samples is insufficient. Semi-MGPPL efficiently reflects interest levels in the latent space by pseudo-labeling of unlabeled samples and increasing the number of available pairs among labeled samples. In addition, obtaining behavior features is difficult for a new test sample. However, requirement of features of all modalities by previous mGPLVM-based methods makes the calculation of latent variables of a test sample challenging. Semi-MGPPL solves this problem by training a projection function from the original feature to the latent space. The experimental results on real data demonstrate the effectiveness and robustness of semi-MGPPL.

This paper presents reliable estimation of deterioration levels via late fusion using multi-view distress images for practical inspection. The proposed method simultaneously solves the following two problems that are necessary to support the practical inspection. Since maintenance of infrastructures requires a high level of safety and reliability, this paper proposes a neural network that can generate an attention map from distress images and text data acquired during the inspection. Thus, deterioration level estimation with high interpretability can be realized. In addition, since multi-view distress images are taken for single distress during the actual inspection, it is necessary to estimate the final result from these images. Therefore, the proposed method integrates estimation results obtained from the multi-view images via the late fusion and can derive an appropriate result considering all the images. To the best of our knowledge, no method has been proposed to solve these problems simultaneously, and this achievement is the biggest contribution of this paper. In this paper, we confirm the effectiveness of the proposed method by conducting experiments using data acquired during the actual inspection.

A few-shot personalized saliency prediction based on adaptive image selection considering object and visual attention is presented in this paper. Since general methods predicting personalized saliency maps (PSMs) need a large number of training images, the establishment of a theory using a small number of training images is needed. To tackle this problem, although finding persons who have visual attention similar to that of a target person is effective, all persons have to commonly gaze at many images. Thus, it becomes difficult and unrealistic when considering their burden. On the other hand, this paper introduces a novel adaptive image selection (AIS) scheme that focuses on the relationship between human visual attention and objects in images. AIS focuses on both a diversity of objects in images and a variance of PSMs for the objects. Specifically, AIS selects images so that selected images have various kinds of objects to maintain their diversity. Moreover, AIS guarantees the high variance of PSMs for persons since it represents the regions that many persons commonly gaze at or do not gaze at. The proposed method enables selecting similar users from a small number of images by selecting images that have high diversities and variances. This is the technical contribution of this paper. Experimental results show the effectiveness of our personalized saliency prediction including the new image selection scheme.

An interpretable convolutional neural network (CNN) including attribute estimation for image classification is presented in this paper. Although CNNs perform highly accurate image classification, the reason for the classification results obtained by the neural networks is not clear. In order to provide interpretation of CNNs, the proposed method estimates attributes, which explain elements of objects, in an intermediate layer of the network. This enables improvement of the interpretability of CNNs, and it is the main contribution of this paper. Furthermore, the proposed method uses the estimated attributes for image classification in order to enhance its accuracy. Consequently, the proposed method not only provides interpretation of CNNs but also realizes improvement in the performance of image classification.

The details of the matches of soccer can be estimated from visual and audio sequences, and they correspond to the occurrence of important scenes. Therefore, the use of these sequences is suitable for important scene detection. In this paper, a new multimodal method for important scene detection from visual and audio sequences in far-view soccer videos based on a single deep neural architecture is presented. A unique point of our method is that multiple classifiers can be realized by a single deep neural architecture that includes a Convolutional Neural Network-based feature extractor and a Support Vector Machine-based classifier. This approach provides a solution to the problem of not being able to simultaneously optimize different multiple deep neural architectures from a small amount of training data. Then we monitor confidence measures output from this architecture for the multimodal data and enable their integration to obtain the final classification result.

The paper proposes a method of visual attention-based emotion classification through eye gaze analysis. Concretely, tensor-based emotional category classification via visual attention-based heterogeneous convolutional neural network (CNN) feature fusion is proposed. Based on the relationship between human emotions and changes in visual attention with time, the proposed method performs new gaze-based image representation that is suitable for reflecting the characteristics of the changes in visual attention with time. Furthermore, since emotions evoked in humans are closely related to objects in images, our method uses a CNN model to obtain CNN features that can represent their characteristics. For improving the representation ability to the emotional categories, we extract multiple CNN features from our novel gaze-based image representation and enable their fusion by constructing a novel tensor consisting of these CNN features. Thus, this tensor construction realizes the visual attention-based heterogeneous CNN feature fusion. This is the main contribution of this paper. Finally, by applying logistic tensor regression with general tensor discriminant analysis to the newly constructed tensor, the emotional category classification becomes feasible. Since experimental results show that the proposed method enables the emotional category classification with the F1-measure of approximately 0.6, and about 10% improvement can be realized compared to comparative methods including state-of-the-art methods, the effectiveness of the proposed method is verified.

A new approach that drastically improves cross-modal retrieval performance in vision and language (hereinafter referred to as & x201C;vision and language retrieval & x201D;) is proposed in this paper. Vision and language retrieval takes data of one modality as a query to retrieve relevant data of another modality, and it enables flexible retrieval across different modalities. Most of the existing methods learn optimal embeddings of visual and lingual information to a single common representation space. However, we argue that the forced embedding optimization results in loss of key information for sentences and images. In this paper, we propose an effective utilization of representation spaces in a simple but robust vision and language retrieval method. The proposed method makes use of multiple individual representation spaces through text-to-image and image-to-text models. Experimental results showed that the proposed approach enhances the performance of existing methods that embed visual and lingual information to a single common representation space.

BACKGROUND The risk of gastric cancer increases in patients withHelicobacter pylori-associated chronic atrophic gastritis (CAG). X-ray examination can evaluate the condition of the stomach, and it can be used for gastric cancer mass screening. However, skilled doctors for interpretation of X-ray examination are decreasing due to the diverse of inspections. AIM To evaluate the effectiveness of stomach regions that are automatically estimated by a deep learning-based model for CAG detection. METHODS We used 815 gastric X-ray images (GXIs) obtained from 815 subjects. The ground truth of this study was the diagnostic results in X-ray and endoscopic examinations. For a part of GXIs for training, the stomach regions are manually annotated. A model for automatic estimation of the stomach regions is trained with the GXIs. For the rest of them, the stomach regions are automatically estimated. Finally, a model for automatic CAG detection is trained with all GXIs for training. RESULTS In the case that the stomach regions were manually annotated for only 10 GXIs and 30 GXIs, the harmonic mean of sensitivity and specificity of CAG detection were 0.955 +/- 0.002 and 0.963 +/- 0.004, respectively. CONCLUSION By estimating stomach regions automatically, our method contributes to the reduction of the workload of manual annotation and the accurate detection of the CAG.

A novel method for player importance prediction from a player network using gaze positions estimated by Long Short-Term Memory (LSTM) in soccer videos is presented in this paper. By newly using an estimation model of gaze positions trained by gaze tracking data of experienced persons, it is expected that the importance of each player can be predicted. First, we generate a player network by utilizing the estimated gaze positions and first-arrival regions representing players' connections, e.g., passes between players. The gaze positions are estimated by LSTM that is newly trained from the gaze tracking data of experienced persons. Second, the proposed method predicts the importance of each player by applying the Hypertext Induced Topic Selection (HITS) algorithm to the constructed network. Consequently, prediction of the importance of each player based on soccer tactic knowledge of experienced persons can be realized without constantly obtaining gaze tracking data.

A method for image retrieval based on supervised local regression and global alignment (sLRGA) with relevance feedback for insect identification is presented in this paper. Based on the novel sLRGA, which is an extended version of LRGA, the proposed method estimates ranking scores for image retrieval in such a way that the neighborhood structure of a feature space of the database can be optimally preserved with consideration of class information. This is the main contribution of this paper. By measuring the relevance between all of the images and the query image in the database, sLRGA realizes accurate image retrieval. Furthermore, when positive/negative labels to retrieved images are given by users, the proposed method can improve image retrieval performance considering the query relevance information via use of both relevance feedback and sLRGA. This is the second contribution of this paper. Experimental results show the effectiveness of the proposed method.

A method for estimating interest levels from behavior features via tensor completion including adaptive similar user selection is presented in this paper. The proposed method focuses on a tensor that is suitable for data containing multiple contexts and constructs a third-order tensor in which three modes are "products", "users" and "user behaviors and interest levels" for these products. By complementing this tensor, unknown interest level estimation of a product for a target user becomes feasible. For further improving the estimation performance, the proposed method adaptively selects similar users for the target user by focusing on converged estimation errors between estimated interest levels and known interest levels in the tensor completion. Furthermore, the proposed method can adaptively estimate the unknown interest from the similar users. This is the main contribution of this paper. Therefore, the influence of users having different interests is reduced, and accurate interest level estimation can be realized. In order to verify the effectiveness of the proposed method, we show experimental results obtained by estimating interest levels of users holding books.

A novel trial for estimating popularity of artists in music streaming services (MSS) is presented in this paper. The main contribution of this paper is to improve extensibility for using multi-modal features to accurately analyze latent relationships between artists. In the proposed method, a novel framework to construct a network is derived by collaboratively using social metadata and multi-modal features via canonical correlation analysis. Different from conventional methods that do not use multi-modal features, the proposed method can construct a network that can capture social metadata and multi-modal features, i.e., a context-aware network. For effectively analyzing the context-aware network, a novel framework to realize popularity estimation of artists is developed based on network analysis. The proposed method enables effective utilization of the network structure by extracting node features via a node embedding algorithm. By constructing an estimator that can distinguish differences between the node features, the proposed method can archive accurate popularity estimation of artists. Experimental results using multiple real-world datasets that contain artists in various genres in Spotify, one of the largest MSS, are presented. Quantitative and qualitative evaluations show that our method is effective for both classifying and regressing the popularity.

Techniques for integrating different types of multiple features effectively have been actively studied in recent years. Multiset canonical correlation analysis (MCCA), which maximizes the sum of pairwise correlations of inter-view (i.e., between different features), is one of the powerful methods for integrating different types of multiple features, and various MCCA-based methods have been proposed. This work focuses on a supervised MCCA variant in order to construct a novel effective feature integration framework. In this paper, we newly propose supervised fractional-order embedding geometrical multi-view CCA (SFGMCCA). This method constructs not only the correlation structure but also two types of geometrical structures of intra-view (i.e., within each feature) and inter-view simultaneously, thereby realizing more precise feature integration. This method also supports the integration of small sample and high-dimensional data by using the fractional-order technique. We conducted experiments using four types of image datasets, i.e., MNIST, COIL-20, ETH-80 and CIFAR-10. Furthermore, we also performed an fMRI dataset containing brain signals to verify the robustness. As a result, it was confirmed that accuracy improvements using SFGMCCA were statistically significant at the significance level of 0.05 compared to those using conventional representative MCCA-based methods.

High-quality annotations for medical images are always costly and scarce. Many applications of deep learning in the field of medical image analysis face the problem of insufficient annotated data. In this paper, we present a semi-supervised learning method for chronic gastritis classification using gastric X-ray images. The proposed semi-supervised learning method based on tri-training can leverage unannotated data to boost the performance that is achieved with a small amount of annotated data. We utilize a novel learning method named Between-Class learning (BC learning) that can considerably enhance the performance of our semi-supervised learning method. As a result, our method can effectively learn from unannotated data and achieve high diagnostic accuracy for chronic gastritis.

Aim: The aim of this study was to determine whether our deep convolutional neural network-based anomaly detection model can distinguish differences in esophagus images and stomach images obtained from gastric X-ray examinations. Methods: A total of 6012 subjects were analyzed as our study subjects. Since the number of esophagus X-ray images is much smaller than the number of gastric X-ray images taken in X-ray examinations, we took an anomaly detection approach to realize the task of organ classification. We constructed a deep autoencoding gaussian mixture model (DAGMM) with a convolutional autoencoder architecture. The trained model can produce an anomaly score for a given test X-ray image. For comparison, the original DAGMM, AnoGAN, and a One-Class Support Vector Machine (OCSVM) that were trained with features obtained by a pre-trained Inception-v3 network were used. Results: Sensitivity, specificity, and the calculated harmonic mean of the proposed method were 0.956, 0.980, and 0.968, respectively. Those of the original DAGMM were 0.932, 0.883, and 0.907, respectively. Those of AnoGAN were 0.835, 0.833, and 0.834, respectively, and those of OCSVM were 0.932, 0.935, and 0.934, respectively. Experimental results showed the effectiveness of the proposed method for an organ classification task. Conclusion: Our deep convolutional neural network-based anomaly detection model has shown the potential for clinical use in organ classification.

A deterioration level estimation method via neural network maximizing category-based ordinally supervised multi-view canonical correlation is presented in this paper. This paper focuses on real world data such as industrial applications and has two contributions. First, a novel neural network handling multi-modal features transforms original features into features effectively representing deterioration levels in transmission towers, which are one of the infrastructures, with consideration of only correlation maximization. It can be realized by setting projection matrices maximizing correlations between multiple features into weights of hidden layers. That is, since the proposed network has only a few hidden layers, it can be trained from a small amount of training data. Second, since there exist diverse characteristics and an ordinal scale in deterioration levels, the proposed method newly derives category-based ordinally supervised multi-view canonical correlation analysis (Co-sMVCCA). Co-sMVCCA enables estimation of effective projection considering both within-class divergence and the ordinal scale between classes. Experimental results showed that the proposed method realizes accurate deterioration level estimation.

In this paper, a novel synthetic gastritis image generation method based on a generative adversarial network (GAN) model is presented. Sharing medical image data is a crucial issue for realizing diagnostic supporting systems. However, it is still dif cult for researchers to obtain medical image data since the data include individual information. Recently proposed GAN models can learn the distribution of training images without seeing real image data, and individual information can be completely anonymized by generated images. If generated images can be used as training images in medical image classi cation, promoting medical image analysis will become feasible. In this paper, we targeted gastritis, which is a risk factor for gastric cancer and can be diagnosed by gastric X-ray images. Instead of collecting a large amount of gastric X-ray image data, an image generation approach was adopted in our method.We newly propose loss function-based conditional progressive growing generative adversarial network (LC-PGGAN), a gastritis image generation method that can be used for a gastritis classi cation problem. The LC-PGGAN gradually learns the characteristics of gastritis in gastric X-ray images by adding new layers during the training step. Moreover, the LC-PGGAN employs loss function-based conditional adversarial learning so that generated images can be used as the gastritis classi cation task. We show that images generated by the LC-PGGAN are effective for gastritis classi cation using gastric X-ray images and have clinical characteristics of the target symptom.

In this paper, we propose a novel scene retrieval and re-ranking method based on a text-toimage Generative Adversarial Network (GAN). The proposed method generates an image from an input query sentence based on the text-to-image GAN and then retrieves a scene that is the most similar to the generated image. By utilizing the image generated from the input query sentence as a query, we can control semantic information of the query image at the text level. Furthermore, we introduce a novel interactive reranking scheme to our retrieval method. Specifically, users can consider the importance of each word within the first input query sentence. Then the proposed method re-generates the query image that reflects the word importance provided by users. By updating the generated query image based on the word importance, it becomes feasible for users to revise retrieval results through this re-ranking process. In experiments, we showed that our retrieval method including the re-ranking scheme outperforms recently proposed retrieval methods.

Background Deep learning has become a new trend of image recognition tasks in the field of medicine. We developed an automated gastritis detection system using double-contrast upper gastrointestinal barium X-ray radiography. Methods A total of 6520 gastric X-ray images obtained from 815 subjects were analyzed. We designed a deep convolutional neural network (DCNN)-based gastritis detection scheme and evaluated the effectiveness of our method. The detection performance of our method was compared with that of ABC (D) stratification. Results Sensitivity, specificity, and harmonic mean of sensitivity and specificity of our method were 0.962, 0.983, and 0.972, respectively, and those of ABC (D) stratification were 0.925, 0.998, and 0.960, respectively. Although there were 18 false negative cases in ABC (D) stratification, 14 of those 18 cases were correctly classified into the positive group by our method.

This paper presents a convolutional sparse coding (CSC)-based deep random vector functional link network (CSDRN) for distress classification of road structures. The main contribution of this paper is the introduction of CSC into a feature extraction scheme in the distress classification. CSC can extract visual features representing characteristics of target images because it can successfully estimate optimal convolutional dictionary filters and sparse features as visual features by training from a small number of distress images. The optimal dictionaries trained from distress images have basic components of visual characteristics such as edge and line information of distress images. Furthermore, sparse feature maps estimated on the basis of the dictionaries represent both strength of the basic components and location information of regions having their components, and these maps can represent distress images. That is, sparse feature maps can extract key components from distress images that have diverse visual characteristics. Therefore, CSC-based feature extraction is effective for training from a limited number of distress images that have diverse visual characteristics. The construction of a novel neural network, CSDRN, by the use of a combination of CSC-based feature extraction and the DRN classifier, which can also be trained from a small dataset, is shown in this paper. Accurate distress classification is realized via the CSDRN.

Scene retrieval from input descriptions has been one of the most important applications with the increasing number of videos on the Web. However, this is still a challenging task since semantic gaps between features of texts and videos exist. In this paper, we try to solve this problem by utilizing a text-to-image Generative Adversarial Network (GAN), which has become one of the most attractive research topics in recent years. The text-to-image GAN is a deep learning model that can generate images from their corresponding descriptions. We propose a new retrieval framework“, Query is GAN”, based on the text-to-image GAN that drastically improves scene retrieval performance by simple procedures. Our novel idea makes use of images generated by the text-to-image GAN as queries for the scene retrieval task. In addition, unlike many studies on text-to-image GANs that mainly focused on the generation of high-quality images, we reveal that the generated images have reasonable visual features suitable for the queries even though they are not visually pleasant. We show the effectiveness of the proposed framework through experimental evaluation in which scene retrieval is performed from real video datasets.

A novel method for estimating team tactics in soccer videos based on a Deep Extreme Learning Machine (DELM) and unique characteristics of tactics is presented in this paper. The proposed method estimates the tactics of each team from players formations and enables successful training from a limited amount of training data. Specifically, the estimation of tactics consists of two stages. First, by utilizing two DELMs corresponding to the two teams, the proposed method estimates the provisional tactics of each team. Second, the proposed method updates the team tactics based on unique characteristics of soccer tactics, the relationship between tactics of the two teams and information on ball possession. Consequently, since the proposed method estimates the team tactics that satisfy these characteristics, accurate estimation results can be obtained. In an experiment, the proposed method is applied to actual soccer videos to verify its effectiveness.

A novel audio feature projection using Kernel Discriminative Locality Preserving Canonical Correlation Analysis (KDLPCCA)-based correlation with electroencephalogram (EEG) features for favorite music classification is presented in this paper. The projected audio features reflect individual music preference adaptively since they are calculated by considering correlations with the user's EEG signals during listening to musical pieces that the user likes/dislikes via a novel CCA proposed in this paper. The novel CCA, called KDLPCCA, can consider not only a non-linear correlation but also local properties and discriminative information of each class sample, namely, music likes/dislikes. Specifically, local properties reflect intrinsic data structures of the original audio features, and discriminative information enhances the power of the final classification. Hence, the projected audio features have an optimal correlation with individual music preference reflected in the user's EEG signals, adaptively. If the KDLPCCA-based projection that can transform original audio features into novel audio features is calculated once, our method can extract projected audio features from a new musical piece without newly observing individual EEG signals. Our method therefore has a high level of practicability. Consequently, effective classification of user's favorite musical pieces via a Support Vector Machine (SVM) classifier using the new projected audio features becomes feasible. Experimental results show that our method for favorite music classification using projected audio features via the novel CCA outperforms methods using original audio features, EEG features and even audio features projected by other state-of-the-art CCAs.

A method for estimating regions of deterioration in electron microscope images of rubber materials is presented in this paper. Deterioration of rubber materials is caused by molecular cleavage, external force, and heat. An understanding of these characteristics is essential in the eld of material science for the development of durable rubber materials. Rubber material deterioration can be observed by using on electron microscope but it requires much effort and specialized knowledge to nd regions of deterioration. In this paper, we propose an automated deterioration region estimation method based on deep learning and anomaly detection techniques to support such material development. Our anomaly detection model, called Transfer Learning-based Deep Autoencoding Gaussian Mixture Model (TL-DAGMM), uses only normal regions for training since obtaining training data for regions of deterioration is dif cult. TL-DAGMMmakes use of extracted high representation features from a pre-trained deep learning model and can automatically learn the characteristics of normal rubber material regions. Regions of deterioration are estimated at the pixel level by calculated anomaly scores. Experiments on real rubber material electron microscope images demonstrated the effectiveness of our model.

This paper presents a language-independent ontology (LION) construction method that uses tagged images in an image folksonomy. Existing multilingual frameworks that construct an ontology deal with concepts translated on the basis of parallel corpora, which are not always available; however, the proposed method enables LION construction without parallel corpora by using visual features extracted from tagged images as the alternative. In the proposed method, visual similarities in tagged images are leveraged to aggregate synonymous concepts across languages. The aggregated concepts take on intrinsic semantics of themselves, while they also hold distinct characteristics in different languages. Then relationships between concepts are extracted on the basis of visual and textual features. The proposed method constructs a LION whose nodes and edges correspond to the aggregated concepts and relationships between them, respectively. The LION enables successful image retrieval across languages since each of the aggregated concepts can be referred to in different languages. Consequently, the proposed method removes the language barriers by providing an easy way to access a broader range of tagged images for users in the folksonomy, regardless of the language they use.

Video reranking is an effective way for improving the retrieval performance of text-based video search engines. This paper proposes a graph-based Web video search reranking method with local and global consistency analysis. Generally, the graph-based reranking approach constructs a graph whose nodes and edges respectively correspond to videos and their pairwise similarities. A lot of reranking methods are built based on a scheme which regularizes the smoothness of pairwise relevance scores between adjacent nodes with regard to a user’s query. However, since the overall consistency is measured by aggregating only the local consistency over each pair, errors in score estimation increase when noisy samples are included within query-relevant videos’ neighbors. To deal with the noisy samples, the proposed method leverages the global consistency of the graph structure, which is different from the conventional methods. Specifically, in order to detect this consistency, the propose method introduces a spectral clustering algorithm which can detect video groups, in which videos have strong semantic correlation, on the graph. Furthermore, a new regularization term, which smooths ranking scores within the same group, is introduced to the reranking framework. Since the score regularization is performed by both local and global aspects simultaneously, the accurate score estimation becomes feasible. Experimental results obtained by applying the proposed method to a real-world video collection show its effectiveness.

This paper presents estimation of deterioration levels of transmission towers via deep learning maximizing the canonical correlation between heterogeneous features. In the proposed method, we newly construct a correlation-maximizing deep extreme learning machine based on a local receptive field (CMDELM-LRF). For accurate deterioration level estimation, it is necessary to obtain semantic information that effectively represents deterioration levels. However, since the amount of training data for transmission towers is small, it is difficult to perform feature transformation by using many hidden layers such as general deep learning methods. In CMDELM-LRF, one hidden layer, which maximizes the canonical correlation between visual features and text features obtained from inspection text data, is newly inserted. Specifically, by using projections obtained by maximizing the canonical correlation as weight parameters of the hidden layer, feature transformation for extracting semantic information is realized without designing many hidden layers. This is the main contribution of this paper. Consequently, CMDELM-LRF realizes accurate deterioration level estimation from a small amount of training data.

This paper presents a method for automatic detection of fish sounds in an underwater environment. There exist two difficulties: (i) features and classifiers that provide good detection results differ depending on the underwater environment and (ii) there are cases where a large amount of training data that is necessary for supervised machine learning cannot be prepared. A method presented in this paper (the proposed hybrid method) overcomes these difficulties as follows. First, novel logistic regression (NLR) is derived via adaptive feature weighting by focusing on the accuracy of classification results by multiple classifiers, support vector machine (SVM), and k-nearest neighbors (kNN). Although there are cases where SVM or k-NN cannot work well due to divergence of useful features, NLR can produce complementary results. Second, the proposed hybrid method performs multi-stage classification with consideration of the accuracy of SVM, k-NN, and NLR. The multistage acquisition of reliable results works adaptively according to the underwater environment to reduce performance degradation due to diversity of useful classifiers even if abundant training data cannot be prepared. Experiments on underwater recordings including sounds of Sciaenidae such as silver croakers (Pennahia argentata) and blue drums (Nibea mitsukurii) show the effectiveness of the proposed hybrid method.