Abstract LiDAR odometry estimation and 3D semantic segmentation are crucial for autonomous driving, which has achieved remarkable advances recently. However, these tasks are challenging due to the imbalance of points in different semantic categories for 3D semantic segmentation and the influence of dynamic objects for LiDAR odometry estimation, which increases the importance of using representative/salient landmarks as reference points for robust feature learning. To address these challenges, we propose a saliency-guided approach that leverages attention information to improve the performance of LiDAR odometry estimation and semantic segmentation models. Unlike in the image domain, only a few studies have addressed point cloud saliency information due to the lack of annotated training data. To alleviate this, we first present a universal framework to transfer saliency distribution knowledge from color images to point clouds, and use this to construct a pseudo-saliency dataset (i.e. FordSaliency) for point clouds. Then, we adopt point cloud based backbones to learn saliency distribution from pseudo-saliency labels, which is followed by our proposed SalLiDAR module. SalLiDAR is a saliency-guided 3D semantic segmentation model that integrates saliency information to improve segmentation performance. Finally, we introduce SalLONet, a self-supervised saliency-guided LiDAR odometry network that uses the semantic and saliency predictions of SalLiDAR to achieve better odometry estimation. Our extensive experiments on benchmark datasets demonstrate that the proposed SalLiDAR and SalLONet models achieve state-of-the-art performance against existing methods, highlighting the effectiveness of image-to-LiDAR saliency knowledge transfer. Source code will be available at https://github.com/nevrez/SalLONet

We introduce Mixup Gamblers+, a method for building a deep learning classifier capable of self-evaluating the reliability of inference results. In the proposed method, samples with high uncertainty are generated virtually through data interpolation in the feature space embedded by the deep learning model, and the classifier is trained to detect them. Moreover, we introduce metric learning for feature representation learning to correlate the distance in the latent feature space with the similarity of the samples. This enables us to estimate the uncertainty of the pseudo data based on the distance in the latent feature space and enhances the training of our proposed method. The use of data interpolation in the latent feature space makes the proposed method a general-purpose method for learning uncertainty estimation models for inference, independent of the dataset and problem settings. The proposed method improves the accuracy of the inference and achieves state-of-the-art inferential uncertainty estimation results.

The use of computer-assisted diagnosis (CAD) systems for pathological image analysis constitutes an important research topic. Such systems should be accurate, and their decisions should be explainable to ensure reliability. In this paper, we present an explainable diagnosis method based on convolutional neural networks (CNNs). This method allows us to interpret the basis of the decisions made by the CNN from two perspectives, namely statistics and visualization. For the statistical explanation, the method constructs dictionaries of representative pathological features over multiple scales in training data. It performs diagnoses based on the occurrence and importance of items in the dictionaries to rationalize its decisions. We introduce a vector quantization scheme to the CNN to enable it to construct the feature dictionary. For the visual interpretation, the method provides images of learned features in the dictionary by decoding them from a high-dimensional feature space to a pathological image space. The experimental results showed that the proposed network learned pathological features, which contributed to the diagnosis, and the method yielded approximately an area under the receiver operating curve (AUC) of 0.89 for detecting atypical tissues in pathological images of a uterine cervix by using these features.

The development of computer-assisted diagnosis (CAD) algorithms for pathological image analysis is an important research topic. Recently, convolutional neural networks (CNNs) have been used in several studies for the development of CAD algorithms. Such systems are required to be not only accurate but also explainable for their decisions, to ensure reliability. However, a limitation of using CNNs is that the basis of the decisions made by them are hardly interpretable by humans. Thus, in this paper, we present an explainable diagnosis method comprising two CNNs playing different roles. This method allows us to interpret the basis of the decisions made by the CNN from two perspectives: statistics and visualization. For the statistical explanation, the method constructs a dictionary of representative pathological features. It performs diagnoses based on the occurrence and importance of learned features referred from its dictionary. To construct the dictionary, we introduce a vector quantization scheme for CNN. For the visual interpretation, the method provides images of learned features embedded in a feature space as an index of the dictionary by generating them using a conditional autoregressive model. The experimental results showed that the proposed network learned pathological features that contributed to the diagnosis and yielded an area under the receiver operating curve (AUC) of approximately 0.93 for detecting atypical tissues in pathological images of the uterine cervix. Moreover, the proposed method demonstrated that it could provide visually interpretable images to show the rationales behind its decisions. Thus, the proposed method can serve as a valuable tool for pathological image analysis in terms of both its accuracy and explainability.

This paper presents an efficient object detection method from satellite imagery. Among a number of machine learning algorithms, we proposed a combination of two convolutional neural networks (CNN) aimed at high precision and high recall, respectively. We validated our models using golf courses as target objects. The proposed deep learning method demonstrated higher accuracy than previous object identification methods.

The Earth observation satellites have been monitoring the earth's surface for a long time, and the images taken by the satellites contain large amounts of valuable data. However, it is extremely hard work to manually analyze such huge data. Thus, a method of automatic object detection is needed for satellite images to facilitate efficient data analyses. This paper describes a new image feature extended from higher-order local autocorrelation to the object detection of multispectral satellite images. The feature has been extended to extract spectral inter-relationships in addition to spatial relationships to fully exploit multispectral information. The results of experiments with object detection tasks conducted to evaluate the effectiveness of the proposed feature extension indicate that the feature realized a higher performance compared to existing methods.

Solar power generation has attracted significant attention recently as a safe and environmentally friendly renewable energy source. However, generally speaking, since the service lives of solar power systems are relatively long, and since it is difficult to detect anomalies in individual solar panels, such plants tend to operate without much consideration for individual panel anomalies. In order to more comprehensively monitor solar power generation systems, the National Institute of Advanced Industrial Science and Technology (AIST) of Japan has developed a direct current (DC) power line communication system that enables monitoring of each panel in a system. Monitored data are then integrated and uploaded to the cloud. Using this monitored data, we found that the integrated power ratio trends of single panel power output are in accordance with a normal distribution. Therefore, herein, we propose an anomaly detection method that uses a normal distribution. We then describe an experiment using 24 solar panels into which pseudo-faults were induced and show that our proposal makes it possible to detect errors with high accuracy.

Semantic labelling of acoustic scenes has recently emerged as active topic covering a wide range of applications, e.g. surveillance and audio-based information retrieval. In this paper, we present an effective approach for acoustic scene classification through characterizing both background sound textures and acoustic events. The work takes inspiration from the psychoacoustic definition of acoustic scenes, that is, "skeleton of (acoustic) events on a bed of (sound) texture". In detail, we firstly employ distinct models to exploit sound textures and events in acoustic scenes, individually. Subsequently, based on fact that the perceptual importance of two parts will vary with respect to different scene categories, we develop favourable class-conditional fusion scheme to aggregate two-channel information. To validate proposed approach, we conduct extensive experiments on Rouen dataset which includes 19 categories of daily acoustic scenes with 3026 real-world recordings, and the proposed approach outperforms state-of-the-art methods by a large margin.

In this paper, we present a novel method for environmental sound classification in non-stationary noise environment. The proposed method mainly consists of three stages: noise source separation and acoustic feature extraction and multi-class classification. At first stage, we employ probabilistic latent component analysis (PLCA) to perform time-varying noise separation. To alleviate the artifacts introduced by source separation, a series of spectral weightings is applied to enhance reliability of audio spectra. At feature extraction stage, we extract acoustic subspace to effectively characterize temporal-spectral patterns of denoised sound spectrogram. Subsequently, regularized kernel Fisher discriminant analysis (KFDA) is adopted to conduct multi-class sound classification through exploiting class conditional distributions based on extracted acoustic subspaces (features). The proposed method is evaluated with Real World Computing Partnership (RWCP) sound scene database and experimental results demonstrate its superior performance compared to other methods.

In this paper, we propose an effective discriminant subspace learning framework to recognize the environmental sounds. Firstly, Gabor transform is adopted to characterize the time-frequency distributions of environmental sounds. We further encode the prominent time-frequency patterns with low rank representation by extracting the subspace from Gabor spectrogram. Unlike conventional sound recognition schemes that are mostly based on acoustic feature vectors, we treat the acoustic subspaces (matrixes) as basic elements for recognition, retaining rich temporal-spectral contextual information. At recognition stage, we employ kernel Fisher discriminant analysis to effectively exploit the class conditional distributions of environmental sounds which are favorable for performing multi-class classification. With a well developed kernel function, the proposed approach achieved superior recognition performance on RWCP sound scene database, compared with the existing methods.

Capsule endoscopy is a painless way more and more utilized in gastrointestinal examination. Nevertheless, there is an issue comes out that the efficiency and accuracy of capsule endoscopy diagnosis is now restricted by the large quantity of images. In this paper, an anomaly detection method for capsule endoscopy images captured within the range of small intestine is described. Aiming to realize the anomaly detection, this paper takes the advantage of Higher-order Local Auto Correlation features and subspace method using PCA (Principal Component Analysis). The proposed method is validated over capsule endoscopy image sets and its effectiveness is demonstrated by experimental results.

To solve the problem of fluctuations in clock timing with digital LSIs (also known as the "clock skew" problem), we propose a genetic algorithm (GA) based clock adjustment method that ensures robust clock-timing to cope with fluctuations in the LSI environment such as temperature or power supply voltage. This method is realized by the combination of dedicated adjustable circuitry and adjustment GA software, with the values for multiple adjustable delay circuits inserted into the clock lines being determined by the GA software after fabrication. Experimental results demonstrate that the proposed method can enhance the operational yields of developed test chips while ensuring sufficient timing margins.

This paper presents the first successful application of multi-objective genetic algorithms (MOGA) to transmission line modeling. Conventionally, it is difficult to simultaneously realize high-accuracy transmission line simulations for both the frequency and time-domains. In order to overcome the problem, this paper proposes the application of MOGA to transmission line modeling. The proposed modeling method has two distinctive features: (1) Simultaneous modeling of both frequency and time-domain characteristics. (2) Selection of conventionalmodel from among set of Pareto optimization solutions based on simulation objectives. The results of an experiment with a micro-strip line demonstrate that the proposed method accomplished a transmission line simulation up to 2.54 times more accurate than conventional methods.

In this paper, we present an automatic parameter extraction method with the GA (Genetic Algorithm) for surface-potential-based MOSFET models such as HiSIM (Hiroshima-university STARC IGFET Model). The method employs a two-stage extraction procedure operating on different sets of model parameters. Experimental results demonstrate that extraction of 34 parameters can be completed within 23 hours with PC (AthlonXP 2500), although this would typically take a human expert several days.

This paper describes an automatic multi-objective adjustment system for optical axes using genetic algorithms. It is difficult for conventional systems to automatically adjust optical axes, because it requires high-precision positioning and angle setting with /spl mu/m resolution. Moreover, multiple goals that have a trade-off relation must be satisfied simultaneously by the adjustment. In order to overcome this problem, we propose a multi-objective adjustment system using genetic algorithms. In experiments, simultaneous alignment for the positioning and the angles (parallelism) of optical axes, which is difficult with conventional methods, could be realized within three hours.

This paper describes the automatic adjustment of a femtosecond-pulse laser using genetic algorithms (GA). Laser systems must be precisely adjusted, because the light beam has to travel many times within a laser cavity before returning to the focus point with submicron resolution. Therefore, it typically takes three days or more to adjust a femtosecond laser manually, requiring much trial and error to determine the optimal settings. In order to overcome this problem, we propose an automatic adjustment method using genetic algorithms to quickly achieve the optimal settings for a femtosecond-pulse laser. We have also developed new mirror holders allowing for both high-precision and high-speed adjustment. With this laser system, the optical components can be automatically adjusted within 15 minutes.

This paper describes the GRD (Genetic Reconfiguration of DSPs) chip, which is evolvable hardware designed for neural network applications. The GRD chip is a building block for the configuration of a scalable neural network hardware system. Both the topology and the hidden layer node functions of a neural network mapped on the GRD chips are dynamically reconfigured using a genetic algorithm (GA). Thus, the most desirable network topology and choice of node functions (e.g., Gaussian or sigmoid function) for a given application can be determined adaptively. This approach is particularly suited to applications requiring the ability to cope with time-varying problems and real-time constraints. The GRD chip consists of a 100 MHz 32-bit RISC processor and 15 33 MHz 16-bit DSPs connected in a binary-tree network. The RISC processor is the NEC V830 which executes mainly the GA. According to chromosomes obtained by the GA, DSP functions and the interconnection among them are dynamically reconfigured. The GRD chip does not need a host machine for this reconfiguration. This is desirable for embedded systems in practical industrial applications. Simulation results on chaotic time series prediction are two orders of magnitude faster than on a Sun Ultra 2.

In contrast to conventional hardware where the structure is irreversibly fixed in the design process, evolvable hardware (EHW) is designed to adapt to changes in task requirements or changes in the environment, through its ability to reconfigure its own hardware structure dynamically and autonomously. This capacity for adaptation, achieved by employing efficient search algorithms based on the metaphor of evolution, has great potential for the development of innovative industrial applications. This paper introduces EHW chips and six applications currently being developed as part of MITI's Real-World Computing Project; an analog EHW chip for cellular phones, a clock-timing architecture for Giga hertz systems, a neural network EHW chip capable of autonomous reconfiguration, a data compression EHW chip for electrophotographic printers, and a gate-level EHW chip for use in prosthetic hands and robot navigation.