Although energy conservation is in demand worldwide, energy efficiency has not been given much consideration in batch processes where high-value-added products are produced in. However, even in batch processes, energy saving needs to be taken into consideration. In the chemical plants, direct heat recovery between hot and cold streams is one of the major technologies to achieve energy conservation. This technology has been introduced mainly in continuous processes such as petrochemical production. And it is widely used in current continuous plants. However, it is difficult to implement in batch processes because there must be both hot and cold streams at the same time for direct heat recovery. Therefore, I have developed a scheduling system to consider heat recovery in batch processes. In this system, the schedules are generated using genetic algorithm. In this study, the performances of the schedules are compared from the perspective of the "selection" and/or "crossover" methods of the genetic algorithm.

Speech is inappropriate in many situations, limiting when voice control can be used. Most unvoiced speech text entry systems can not be used while on-the-go due to movement artifacts. Using a dental retainer with capacitive touch sensors, SilentSpeller tracks tongue movement, enabling users to type by spelling words without voicing. SilentSpeller achieves an average 97% character accuracy in offline isolated word testing on a 1164-word dictionary. Walking has little effect on accuracy; average offline character accuracy was roughly equivalent on 107 phrases entered while walking (97.5%) or seated (96.5%). To demonstrate extensibility, the system was tested on 100 unseen words, leading to an average 94% accuracy. Live text entry speeds for seven participants averaged 37 words per minute at 87% accuracy. Comparing silent spelling to current practice suggests that SilentSpeller may be a viable alternative for silent mobile text entry.

This article presents SSR7000, a corpus of synchronized ultrasound tongue and lip images designed for end-to-end silent speech recognition (SSR). Although neural end-to-end models are successfully updating the state-of-the-art technology in the field of automatic speech recognition, SSR research based on ultrasound tongue imaging has still not evolved past cascaded DNN-HMM models due to the absence of a large dataset. In this study, we constructed a large dataset, namely SSR7000, to exploit the performance of the end-to-end models. The SSR7000 dataset contains ultrasound tongue and lip images of 7484 utterances by a single speaker. It contains more utterances per person than any other SSR corpus based on ultrasound imaging. We also describe preprocessing techniques to tackle data variances that are inevitable when collecting a large dataset and present benchmark results using an end-to-end model. The SSR7000 corpus is publicly available under the CC BY-NC 4.0 license.

With the increased use of smart speakers, silent speech interaction (SSI) is attracting attention. Unfortunately, traditional silent speech interaction methods require the addition of obtrusive sensors and devices around the user's face, making wearability and portability a challenge. Considering that most uses for smart speakers do not require many words, we suggest a more casual approach, TieLent, which can easily be worn between the neck and the chest. TieLent's RGB camera is set away from the user's face, presenting less interference with the user. Although TieLent's camera is not able to capture the whole mouth, when combined with our image-to-speech neural network model, it is able to generate the recognizable speech of 15 commands with an average accuracy of 94%.

One of the most difficult things in practicing musical instruments is improving timbre. Unlike pitch and rhythm, timbre is a high-dimensional and sensuous concept, and learners cannot evaluate their timbre by themselves. To efficiently improve their timbre control, learners generally need a teacher to provide feedback about timbre. However, hiring teachers is often expensive and sometimes difficult. Our goal is to develop a low-cost learning system that substitutes the teacher. We found that a variational autoencoder (VAE), which is an unsupervised neural network model, provides a 2-dimensional user-friendly mapping of timbre. Our system, SonoSpace, maps the learner's timbre into a 2D latent space extracted from an advanced player's performance. Seeing this 2D latent space, the learner can visually grasp the relative distance between their timbre and that of the advanced player. Although our system was evaluated mainly with an alto saxophone, SonoSpace could also be applied to other instruments, such as trumpets, flutes, and drums.

The availability of digital devices operated by voice is expanding rapidly. However, the applications of voice interfaces are still restricted. For example, speaking in public places becomes an annoyance to the surrounding people, and secret information should not be uttered. Environmental noise may reduce the accuracy of speech recognition. To address these limitations, a system to detect a user's unvoiced utterance is proposed. From internal information observed by an ultrasonic imaging sensor attached to the underside of the jaw, our proposed system recognizes the utterance contents without the user's uttering voice. Our proposed deep neural network model is used to obtain acoustic features from a sequence of ultrasound images. We confirmed that audio signals generated by our system can control the existing smart speakers. We also observed that a user can adjust their oral movement to learn and improve the accuracy of their voice recognition.

We conduct a two-part study to better understand pen grip postures for general input like mode switching and com-mand invocation. The first part of the study asks participants what variations of their normal pen grip posture they might use, without any specific consideration for sensing capabilities. The second part evaluates three of their sug-gested postures with an additional set of six postures designed for the sensing capabilities of a consumer EMG armband. Results show that grips considered normal and mature, such as the dynamic tripod and the dynamic quadrupod, are the best candidates for pen-grip based interaction, followed by finger-on-pen postures and grips using pen tilt. A convolutional neural network trained on EMG data gathered during the study yields above 70% within-participant recognition accuracy for common sets of five postures and above 80% for three-posture subsets. Based on the results, we propose design guidelines for pen interaction using variations of grip postures.

We propose a system, called ExtVision, to augment visual experiences by generating and projecting context-images onto the periphery of the television or computer screen. A peripheral projection of the context-image is one of the most effective techniques to enhance visual experiences. However, the projection is not commonly used at present, because of the difficulty in preparing the context-image. In this paper, we propose a deep neural network-based method to generate context-images for peripheral projection. A user study was performed to investigate the manner in which the proposed system augments traditional visual experiences. In addition, we present applications and future prospects of the developed system.

Immersion is an important factor in video experiences. Therefore, various methods and video viewing systems have been proposed so far. Although head-mounted displays (HMDs) are home-friendly and more available among these devices, they can provide an immersive video experience owing to their wide field-of-view (FoV) and separation of users from the outside environment. They are often used for panoramic and stereoscopic VR videos, but the demand for viewing standard plane videos has increased in recent years. However, the theater mode, which restricts the FoV, is basically used for viewing plane videos. Thus, the advantages of HMDs are not fully utilized. Therefore, we explored an effective method for viewing plane videos by an HMD, in combination with view augmentation by LED implants to the HMD. We used deep neural network (DNN) to generate images for peripheral vision and wide FoV customization.

Support is required for operator activity in the correcting abnormalities in chemical plants. A plant alarm system must provide useful information to operators as the third layer of the Independent Protection Layers. Therefore, a method for designing a plant alarm system is important for plant safety. Because plants are modified throughout their plant lifecycles, any alarm systems need to be properly managed throughout the plant lifecycles. To manage the changes, the design rationale of the alarm system should be explained explicitly. Takeda et al. (2013) [3] proposed a logical and systematic alarm system design method that explicitly explains design rationales from know-why information for appropriate management of change throughout the plant lifecycle. For the combined or branched component of a cause-effect (CE) model, multiple alternative modules have been proposed. We propose a method of generating alternative modules for a plant alarm system based on first-out alarm alternative signals for the combined or branched component of a CE model.

A map is one of the most useful media in disseminating urgent location-based information (e.g., information on urgent events like congestion, accidents, fire, or sales within a limited time). However, it is generally difficult to recognize the geographical features of urgent information by utilizing maps in mobile computing environments, because there is no technology that automatically focuses on areas within the small display frame of the mobile tool being used.In this paper, we present an implementation method of a location-based active map transformation system that automatically focuses on areas in the display frame according to urgent location-based information being transmitted. Its main feature is that it pulls urgent information into the frame while maintaining the quality of the geographical information being represented. The new strategy reduces the number of new objects in the display frame and the number of objects deleted from it. As a result, it provides a map mobile users can understand.We clarify the feasibility and effectiveness of this method through several experiments.

We propose and examine methods for analyzing an EEG (electroencephalogram) during music listening in particular based on information flow combined with the source derivation (SD) method. The relationship between EEG and music cognition is also discussed. Music can be regarded as input to the brain system which influences human mentality. Since music cognition has many emotional aspects, it is expected that the EEG recorded during music listening may reflect the electrical activity of brain regions related to those emotional aspects. The brain system consists of dozens of functional subsystems and we agree with the hypothesis that their cooperative process enables high level recognition such as music listening. From the point of view of informational neurobiology, we extracted the spatial and temporal frequency patterns from the EEG. First, the SD method was applied to the raw data of EEGs as pre-processing. The SD method is a powerful technique to avoid noises which are evoked and overlapped from other regions of the brain. We improved this method for realizing higher accuracy. Then, we calculated the information flow between EEG channels by directional coherence analysis. In this paper, we especially focused on the directional coherence analysis and spatial patterns of the brain's activated area in response to the music structure. Several observations have shown that there were differences in activated areas between relaxing and music listening conditions.