Conversational interfaces have emerged as an accessible and user-friendly alternative to traditional touch-based self-service kiosks in food-ordering systems. Despite their promise, building such systems remains challenging due to the need for costly data annotation, store-specific model adaptation, and scalable deployment. In this study, we propose a fully automated, end-to-end framework that transforms structured menu databases into high-quality annotated datasets and efficiently deploys store-specific conversational models using a parameter-efficient fine-tuning method. Our approach fine-tunes only 0.9% of the backbone model parameters per store, enabling cost-effective and plug-and-play deployment across diverse environments. To enhance robustness, we further integrate a recommendation module that suggests alternative items when requested menu options are unavailable. Experimental results on data from 27 stores in South Korea demonstrate that our framework consistently outperforms existing data generation baselines in intent classification and slot filling performance, while maintaining high annotation quality. Simulated real-world voice-ordering scenarios confirm the practicality of our framework for rapid, scalable, and accessible deployment in real-world environments.

Large language models (LLMs) have achieved remarkable performance across diverse natural language processing tasks, but concerns persist regarding their potential political biases.While prior research has extensively explored political biases in LLMs' text generation and perception, limited attention has been devoted to biases associated with media outlet names.In this study, we systematically investigate the presence of media outlet name biases in LLMs and evaluate their impact on downstream tasks, such as political bias prediction and news summarization.Our findings demonstrate that LLMs consistently exhibit biases toward the known political leanings of media outlets, with variations across model families and scales.We propose a novel metric to quantify media outlet name biases in LLMs and leverage this metric to develop an automated prompt optimization framework.Our framework effectively mitigates media outlet name biases, offering a scalable approach to enhancing the fairness of LLMs in news-related applications.GitHub

Recent advancements in large language models have significantly influenced the field of online medical consultations.However, critical challenges remain, such as the generation of hallucinated information and the integration of upto-date medical knowledge.To address these issues, we propose Informatics Llama (ILlama), a novel framework that combines retrievalaugmented generation (RAG) with a structured medical knowledge graph.ILlama incorporates relevant medical knowledge by transforming subgraphs from a structured medical knowledge graph into text for RAG.By generating subgraphs from the medical knowledge graph in advance for RAG, specifically focusing on diseases and symptoms, ILlama enhances the accuracy and relevance of its medical reasoning.This framework enables effective incorporation of causal relationships between symptoms and diseases.Also, it delivers context-aware consultations aligned with user queries.Experimental results on the two medical consultation datasets demonstrate that ILlama outperforms strong baselines, achieving a semantic similarity F1 score of 0.884 when compared to ground-truth consultation answers.Furthermore, qualitative analysis of ILlama's responses reveals significant improvements in hallucination reduction and clinical usefulness.These results suggest that ILlama has strong potential as a reliable tool for real-world medical consultation environments.Our implementation is available at:

Despite the remarkable performance of deep models on image recognition tasks, they are known to be susceptible to common corruptions such as blur, noise, and low-resolution. Data augmentation is a conventional way to build a robust model by considering these common corruptions during the training. However, a naive data augmentation scheme may result in a non-specialized model for particular corruptions, as the model tends to learn the averaged distribution among corruptions. To mitigate the issue, we propose a new paradigm of training deep image recognition networks that produce clean-like features from any quality image via an invertible neural architecture. The proposed method consists of two stages. In the first stage, we train an invertible network with only clean images under the recognition objective. In the second stage, its inversion, i.e., the invertible decoder, is attached to a new recognition network and we train this encoder-decoder network using both clean and corrupted images by considering recognition and reconstruction objectives. Our two-stage scheme allows the network to produce clean-like and robust features from any quality images, by reconstructing their clean images via the invertible decoder. We demonstrate the effectiveness of our method on image classification and face recognition tasks.

Humans have the ability to robustly recognize objects with various factors of variations such as nonrigid transformations, background noises, and changes in lighting conditions. However, training deep learning models generally require huge amount of data instances under diverse variations, to ensure its robustness. To alleviate the need of collecting large amount of data and better learn to generalize with scarce data instances, we propose a novel meta-learning method which learns to transfer factors of variations from one class to another, such that it can improve the classification performance on unseen examples. Transferred variations generate virtual samples that augment the feature space of the target class during training, simulating upcoming query samples with similar variations. By sharing the factors of variations across different classes, the model becomes more robust to variations in the unseen examples and tasks using small number of examples per class. We validate our model on multiple benchmark datasets for few-shot classification and face recognition, on which our model significantly improves the performance of the base model, outperforming relevant baselines.

BACKGROUND: While next-generation sequencing (NGS) costs have fallen in recent years, the cost and complexity of computation remain substantial obstacles to the use of NGS in bio-medical care and genomic research. The rapidly increasing amounts of data available from the new high-throughput methods have made data processing infeasible without automated pipelines. The integration of data and analytic resources into workflow systems provides a solution to the problem by simplifying the task of data analysis. RESULTS: To address this challenge, we developed a cloud-based workflow management system, Closha, to provide fast and cost-effective analysis of massive genomic data. We implemented complex workflows making optimal use of high-performance computing clusters. Closha allows users to create multi-step analyses using drag and drop functionality and to modify the parameters of pipeline tools. Users can also import the Galaxy pipelines into Closha. Closha is a hybrid system that enables users to use both analysis programs providing traditional tools and MapReduce-based big data analysis programs simultaneously in a single pipeline. Thus, the execution of analytics algorithms can be parallelized, speeding up the whole process. We also developed a high-speed data transmission solution, KoDS, to transmit a large amount of data at a fast rate. KoDS has a file transfer speed of up to 10 times that of normal FTP and HTTP. The computer hardware for Closha is 660 CPU cores and 800 TB of disk storage, enabling 500 jobs to run at the same time. CONCLUSIONS: Closha is a scalable, cost-effective, and publicly available web service for large-scale genomic data analysis. Closha supports the reliable and highly scalable execution of sequencing analysis workflows in a fully automated manner. Closha provides a user-friendly interface to all genomic scientists to try to derive accurate results from NGS platform data. The Closha cloud server is freely available for use from http://closha.kobic.re.kr/ .

Abstract We propose a novel approach for detecting two kinds of partial blur, defocus and motion blur, by training a deep convolutional neural network. Existing blur detection methods concentrate on designing low‐level features, but those features have difficulty in detecting blur in homogeneous regions without enough textures or edges. To handle such regions, we propose a deep encoder‐decoder network with long residual skip‐connections and multi‐scale reconstruction loss functions to exploit high‐level contextual features as well as low‐level structural features. Another difficulty in partial blur detection is that there are no available datasets with images having both defocus and motion blur together, as most existing approaches concentrate only on either defocus or motion blur. To resolve this issue, we construct a synthetic dataset that consists of complex scenes with both types of blur. Experimental results show that our approach effectively detects and classifies blur, outperforming other state‐of‐the‐art methods. Our method can be used for various applications, such as photo editing, blur magnification, and deblurring.

In multi-class indoor semantic segmentation using RGB-D data, it has been shown that incorporating depth feature into RGB feature is helpful to improve segmentation accuracy. However, previous studies have not fully exploited the potentials of multi-modal feature fusion, e.g., simply concatenating RGB and depth features or averaging RGB and depth score maps. To learn the optimal fusion of multimodal features, this paper presents a novel network that extends the core idea of residual learning to RGB-D semantic segmentation. Our network effectively captures multilevel RGB-D CNN features by including multi-modal feature fusion blocks and multi-level feature refinement blocks. Feature fusion blocks learn residual RGB and depth features and their combinations to fully exploit the complementary characteristics of RGB and depth data. Feature refinement blocks learn the combination of fused features from multiple levels to enable high-resolution prediction. Our network can efficiently train discriminative multi-level features from each modality end-to-end by taking full advantage of skip-connections. Our comprehensive experiments demonstrate that the proposed architecture achieves the state-of-the-art accuracy on two challenging RGB-D indoor datasets, NYUDv2 and SUN RGB-D.

SUMMARY: Deep sequencing of small RNAs has become a routine process in recent years, but no dedicated viewer is as yet available to explore the sequence features simultaneously along with secondary structure and gene expression of microRNA (miRNA). We present a highly interactive application that visualizes the sequence alignment, secondary structure and normalized read counts in synchronous multipanel windows. This helps users to easily examine the relationships between the structure of precursor and the sequences and abundance of final products and thereby will facilitate the studies on miRNA biogenesis and regulation. The project manager handles multiple samples of multiple groups. The read alignment is imported in BAM file format. Implemented features comprise sorting, zooming, highlighting, editing, filtering, saving, exporting, etc. Currently, miRseqViewer supports 84 organisms whose annotation is available at miRBase. AVAILABILITY AND IMPLEMENTATION: miRseqViewer, implemented in Java, is available at https://github.com/insoo078/mirseqviewer or at http://msv.kobic.re.kr. CONTACT: [email protected].

The global telecommunication networks would overcome geography and time differences. Recently, telecommunication equipment and service providers seek to connect the wireless mobile telecommunication networks to the Internet directly. These enhanced wireless mobile networks can result in a reorganized set of conventional functions for better developed services.In order to enable these functions, the enhanced all-IP based SS7 communication network is required. The standards associations such as 3rd Generation Partnership Project (3GPP) and 3GPP2 of mobile communication offer the Internet protocol (IP) multimedia sub-system (IMS) as well as the multi-media domain (MMD) systems. The numerous requirements and issues have been emerged in the changeover process of the wireless signaling system no. 7 (SS7) for all-IP network. The crucial point is to ensure interoperability among heterogeneous networks. In this paper, a function named multi-protocol mobility manager (MPMM) is developed. It can satisfy the requirements and issues required in all-IP networks. The proposed MPMM enables roaming functions between 3GPP and 3GPP2 protocols, can reduce the load of the network, and provides other valuable services. The performance of the system is tested and evaluated. From the results of the performance tests, we believe that MPMM can help improve the networks functionalities.

Considers the nonblocking supervisory control problems of nondeterministic discrete event systems (DESs) which are modeled as nondeterministic automata with /spl epsiv/-moves. Based on language models, the paper presents the necessary and sufficient conditions for the existence of a nonblocking supervisor to achieve a given language specification for a nondeterministic DES. The developed nonblocking supervisor always guarantees the absence of blocked states in a controlled nondeterministic system. Moreover, when the language specification does not satisfy the existence conditions, the paper provides the computational algorithm for finding the supremal language of the specification which satisfies the conditions. Furthermore, the results developed are demonstrated through the example of an assembly work station.

This paper addresses the problem of fault-tolerant robust supervisory control of discrete event systems (DES) with model uncertainty. Necessary and sufficient conditions for the existence of a fault-tolerant robust supervisor are derived. If the conditions are satisfied, then tolerable fault event sequences embedded in the system can be achieved. The results obtained are applied to design, modeling, and control of a workcell in a production line consisting of two gas metal arc welding (GMAW) robots, a sensor, and a conveyor.

In designing distributed databases, data allocation is one of the key design issues. Some advantages of optimal data allocation are reducing cost, increasing performance and availability. However, because of that most of current data allocation techniques have considered only cost and performance, those techniques cannot maintain the availability; and are not proper to the system requiring high-availability. In this paper, we suggested the TCOP (Transaction COmmit Probability) model, which is a new availability model considering multiple path failures occurred during the execution time for distributed transactions. In addition, we proposed a data allocation technique using the genetic algorithms, which can maintain the availability and minimize the processing cost. According to the experimental results, the proposed TCOP model and data allocation technique have shown improvement of the data availability.