A data augmentation and feature extraction method using a variational autoencoder (VAE) for acoustic modeling is described. VAE generative model based on Bayesian learning deep framework. can extract latent values its input variables to generate new information. VAEs are widely used pictures sentences. In this paper, applied speech corpus vector from modeling. First, the size of doubled by encoding extracted original utterances The waveforms have "meanings" waveforms. Therefore, be as...

This paper presents the development of a real-time cloud-based in-vehicle air quality monitoring system that enables prediction current and future cabin quality. The designed provides predictive analytics using machine learning algorithms can measure drivers’ drowsiness fatigue based on presented in car. It consists five sensors level CO2, particulate matter, vehicle speed, temperature, humidity. Data from these were collected stored cloud database. A model multilayer perceptron, support...

Accurate activity recognition plays a major role in smart homes to provide assistance and support for users, especially elderly cognitively impaired people. To realize this task, knowledge-driven approaches are one of the emerging research areas that have shown interesting advantages featu res. However, several limitations been associated with these approaches. The produced models usually incomplete capture all types human activities. This resulted limited ability accurately infer users’...

Berry thinning is a crucial process in table grape cultivation. Such visual features as bunch compactness, form, and berry size are important factors affecting market value. Moreover, sufficient space for each to grow also largely influences the final product's quality, such sugar concentration. requires professional skills, it usually accomplished only by experienced farmers. Furthermore, appropriate period limited two weeks; hence, berry-thinning tasks have led bottleneck terms of...

With the continual evolution of cybersecurity threats, development effective intrusion detection systems is increasingly crucial and challenging. This study tackles these challenges by exploring imbalanced multiclass classification, a common situation in network datasets mirroring real-world scenarios. The paper aims to empirically assess performance diverse classification algorithms managing class distributions. Experiments were conducted using UNSW-NB15 benchmark dataset, comprising ten...

Ultra-wideband radar application for sleep breathing monitoring is hampered by the difficulty of obtaining signals non-stationary subjects. This occurs due to imprecise signal clutter removal and poor body movement algorithms extracting accurate signals. Therefore, this paper proposed a Sleep Breathing Detection Algorithm (SBDA) address challenge. First, SBDA introduces combination variance feature with Discrete Wavelet Transform (DWT) tackle issue method used Daubechies wavelets five levels...

Device-free localization (DFL) has become a hot topic in the paradigm of Internet Things. Traditional methods are focused on locating users with attached wearable devices. This involves privacy concerns and physical discomfort especially to that need wear activate those devices daily. DFL makes use received signal strength indicator (RSSI) characterize user’s location based their influence wireless signals. Existing work utilizes statistical features extracted from However, some may not...

Spoken Document Retrieval (SDR) and Term Detection (STD) have been two of the most intensively investigated topics in spoken document processing research according to establishment SDR STD test collections by Text REtrieval Conference (TREC) NIST. Because Japanese researchers also requires such for STD, we established a working group develop these Special Interest Group -Spoken Language Processing (SIG-SLP) Information Society Japan. The has constructed made available collection SDR, is now...

Internet-of-Things (IoT) devices have rapidly become important in understanding conditions an environment. The sensed data from IoT (or sensor) device generally form a time sequential signal where the values vary with time. This study describes processing using recurrent-based neural network and particularly focuses on two sorts of classification tasks: sound tennis swing motion classification. We will introduce these tasks their evaluation results recurrent networks. experimental show that...

To promote the research and development of an android robot with hospitality, we organized a dialogue competition in which task is to serve customer travel destination recommendation task. The acts as salesperson at agency needs help customers choose their desired tourist destinations. This paper describes setting, software distributed for competition, evaluation procedure, results preliminary final rounds competition.

The advent of smart agriculture has revolutionized and streamlined various manual tasks in grape cultivation, one which is berry thinning. This task necessitates experienced farmers to selectively remove a specific number berries from the working bunch, as guided by remaining bunch. In response, this paper introduces novel real-time edge computing application that automates process counting bunch using single 2D image. proposed employs YOLOv5-based object detection techniques (Jocher, 2021)...

We have held dialogue robot competitions in 2020 and 2022 to compare the performances of interactive robots using an android that closely resembles a human. In 2023, third competition DRC2023 was held. The task designed be more challenging than previous travel agent tasks. Since anyone can now develop system LLMs, participating teams are required effectively uses information about situation on spot (real-time information), which is not handled by ChatGPT other systems. has two rounds,...

The lecture is one of the most valuable genres audiovisual data. Though spoken document processing a promising technology for utilizing in various ways, it difficult to evaluate because evaluation require subjective judgment and/or verification large quantities In this paper, test collection retrieval reported. consists target documents about 2, 700 lectures (604 hours) taken from Corpus Spontaneous Japanese (CSJ), 39 queries, relevant passages each query, and automatic transcription speech...

This paper describes an automatic fluency evaluation of spontaneous speech. Although we regularly observe a variety different disfluencies in speech, focus on two types phenomena, i.e., filled pauses and word fragments. aims to reveal that these have effects speech differently. To this end, conduct series SVM classification experiments the Japanese corpus. The experimental results show features derived from fragments are effective evaluating disfluent especially when combined with prosodic...

Inflorescence trimming is a crucial process to produce high-quality table grapes. It can eliminate nutrient competition in bunch and makes it less vulnerable disease development. After trimming, the remaining part of inflorescence should have target length decided by grape variety. This challenging for novice farmers because time constraint. The farmer needs finish before berries develop. paper proposes novel end-to-end measurement method supporting with augmented reality technology....

This paper describes how semi-supervised learning, called peer collaborative learning (PCL), can be applied to the polyphonic sound event detection (PSED) task, which is one of tasks in Detection and Classification Acoustic Scenes Events (DCASE) challenge. Many deep models have been studied determine what kind events occur where for long a given audio clip. The characteristic PCL used this combination ensemble-based knowledge distillation into sub-networks student-teacher model-based...

HTTP and MQTT messaging protocols widely support the Internet of things (IoT) application. These recent years in-vehicle air quality monitoring system has drawn public's attention. With assist Things technology, passengers will be updated with realtime through web page even in a mobile The customised was deployed on 40 shuttle buses, data have collected over year Universiti Malaysia Perlis (UniMAP). In this study, 11 buses were implemented protocol while rest 29 applied protocol. travelled...