Samsung SDS (South Korea)

The mean absolute percentage error (MAPE) is one of the most widely used measures of forecast accuracy, due to its advantages of scale-independency and interpretability. However, MAPE has the significant disadvantage that it produces infinite or undefined values for zero or close-to-zero actual values. In order to address this issue in MAPE, we propose a new measure of forecast accuracy called the mean arctangent absolute percentage error (MAAPE). MAAPE has been developed through looking at MAPE from a different angle. In essence, MAAPE is a slope as an angle, while MAPE is a slope as a ratio, considering a triangle with adjacent and opposite sides that are equal to an actual value and the difference between the actual and forecast values, respectively. MAAPE inherently preserves the philosophy of MAPE, overcoming the problem of division by zero by using bounded influences for outliers in a fundamental manner through considering the ratio as an angle instead of a slope. The theoretical properties of MAAPE are investigated, and the practical advantages are demonstrated using both simulated and real-life data.

Denoising diffusion probabilistic models (DDPM) have shown remarkable performance in unconditional image generation. However, due to the stochasticity of the generative process in DDPM, it is challenging to generate images with the desired semantics. In this work, we propose Iterative Latent Variable Refinement (ILVR), a method to guide the generative process in DDPM to generate high-quality images based on a given reference image. Here, the refinement of the generative process in DDPM enables a single DDPM to sample images from various sets directed by the reference image. The proposed ILVR method generates high-quality images while controlling the generation. The controllability of our method allows adaptation of a single DDPM without any additional learning in various image generation tasks, such as generation from various downsampling factors, multi-domain image translation, paint-to-image, and editing with scribbles.

We propose automatic contrast-limited adaptive histogram equalization (CLAHE) for image contrast enhancement. We automatically set the clip point for CLAHE based on textureness of a block. Also, we introduce dual gamma correction into CLAHE to achieve contrast enhancement while preserving naturalness. First, we redistribute the histogram of the block in CLAHE based on the dynamic range of each block. Second, we perform dual gamma correction to enhance the luminance, especially in dark regions while reducing over-enhancement artifacts. Since automatic CLAHE adaptively enhances contrast in each block while boosting luminance, it is very effective in enhancing dark images and daylight ones with strong dark shadows. Moreover, automatic CLAHE is computationally efficient, i.e., more than 35 frames/s at 1024 × 682 resolution, due to the independent block processing for contrast enhancement. Experimental results demonstrate that automatic CLAHE with dual gamma correction achieves good performance in contrast enhancement and outperforms state-of-the-art methods in terms of visual quality and quantitative measures.

The carbohydrate response element binding protein (ChREBP), a basic helix-loop-helix/leucine zipper transcription factor, plays a critical role in the control of lipogenesis in the liver. To identify the direct targets of ChREBP on a genome-wide scale and provide more insight into the mechanism by which ChREBP regulates glucose-responsive gene expression, we performed chromatin immunoprecipitation-sequencing and gene expression analysis. We identified 1153 ChREBP binding sites and 783 target genes using the chromatin from HepG2, a human hepatocellular carcinoma cell line. A motif search revealed a refined consensus sequence (CABGTG-nnCnG-nGnSTG) to better represent critical elements of a functional ChREBP binding sequence. Gene ontology analysis shows that ChREBP target genes are particularly associated with lipid, fatty acid and steroid metabolism. In addition, other functional gene clusters related to transport, development and cell motility are significantly enriched. Gene set enrichment analysis reveals that ChREBP target genes are highly correlated with genes regulated by high glucose, providing a functional relevance to the genome-wide binding study. Furthermore, we have demonstrated that ChREBP may function as a transcriptional repressor as well as an activator.

This study examined the effects of the organizational climate maturity on knowledge-management performance, measured in terms of knowledge quality and knowledge-sharing level. Reward, top management support, and IT service quality were investigated as the managerial drivers to positively influence such climate maturity. The hypothesized relationships were tested by the partial least square analysis, with data from 42 organizations in Korea. Findings of the study indicate that a more mature (knowledge friendly) organizational climate is linked to higher knowledge-management performance; reward, top management support, and IT service quality are critical managerial drivers influencing such climate maturity.

Among the potential determinants of consumers’ commitments to online shopping site are information features of the Web site, because online shopping consumers have to base their judgment solely on the product or service information presented on the site. When consumers are satisfied with such information features and perceive clear benefits from their relationships with the site, we can expect them to be more committed to the site. In this study, we investigate the relationship between such determinants and consumers’ commitments to an online shopping site. Results of the online survey with 1,278 Korean customers of online bookstores and ticketing services indicate that information satisfaction and relational benefit are highly predictable of consumers’ commitments to an online shopping site. In addition, we found that information satisfaction is affected most by product information quality, while relational benefit is strongly related to service information quality. These results seem to reflect the consumers’ different perceptual weights to different information contents of the Web sites in forming their Web site perceptions.Request access from your librarian to read this article's full text.

Abstract As the number of online degree programs continues to grow among higher education institutions in the United States, engaging online adult learners to online degree programs is getting more difficult than before. Therefore, this study, situated in a land grant university, investigated the motivational factors that contribute to adult learners’ engagement with online graduate degree programs. Based on 190 sets of survey responses, this quantitative study identified four significant motivational factors (intrinsic motivation, short-term extrinsic motivation, long-term extrinsic motivation, and technological willingness) that contributed to their selection of online programs. Gender differences were found to be influential in intrinsic motivation while age differences could affect learners’ short- and long-term extrinsic motivations. Discussions further focused on the implications of the findings in engaging online adult learners in order to sustain online degree programs in higher education.

Contextualized representations from a pre-trained language model are central to achieve a high performance on downstream NLP task. The pre-trained BERT and A Lite BERT (ALBERT) models can be fine-tuned to give state-of-the-art results in sentence-pair regressions such as semantic textual similarity (STS) and natural language inference (NLI). Although BERT-based models yield the [CLS] token vector as a reasonable sentence embedding, the search for an optimal sentence embedding scheme remains an active research area in computational linguistics. This paper explores on sentence embedding models for BERT and ALBERT. In particular, we take a modified BERT network with siamese and triplet network structures called Sentence-BERT (SBERT) and replace BERT with ALBERT to create Sentence-ALBERT (SALBERT). We also experiment with an outer CNN sentence-embedding network for SBERT and SALBERT. We evaluate performances of all sentence-embedding models considered using the STS and NLI datasets. The empirical results indicate that our CNN architecture improves ALBERT models substantially more than BERT models for STS benchmark. Despite significantly fewer model parameters, ALBERT sentence embedding is highly competitive to BERT in downstream NLP evaluations.

The achievement of the pathologic complete response (pCR) has been considered a metric for the success of neoadjuvant chemotherapy (NAC) and a powerful surrogate indicator of the risk of recurrence and long-term survival. This study aimed to develop a multimodal deep learning model that combined clinical information and pretreatment MR images for predicting pCR to NAC in patients with breast cancer. The retrospective study cohort consisted of 536 patients with invasive breast cancer who underwent pre-operative NAC. We developed a deep learning model to fuse high-dimensional MR image features and the clinical information for the pretreatment prediction of pCR to NAC in breast cancer. The proposed deep learning model trained on all datasets as clinical information, T1-weighted subtraction images, and T2-weighted images shows better performance with area under the curve (AUC) of 0.888 as compared to the model using only clinical information (AUC = 0.827, P < 0.05). Our results demonstrate that the multimodal fusion approach using deep learning with both clinical information and MR images achieve higher prediction performance compared to the deep learning model without the fusion approach. Deep learning could integrate pretreatment MR images with clinical information to improve pCR prediction performance.

Purpose This paper seeks to introduce a six‐sigma based methodology for the SCM domain which was developed and has been used in Samsung. Design/methodology/approach The paper provides a detailed description of how and why a six‐sigma‐based methodology for the SCM domain was developed in Samsung and presents a real industry case to illustrate the usage of the methodology. Findings In Samsung, the effort and investment in synthesizing SCM and six sigma, and developing a unique six‐sigma‐based methodology to improve its SCM operation, have turned out to be fruitful. The Black Belt program has produced highly qualified and talented SCM specialists, who are currently training the methodology to members in their organizations and leading SCM projects. SCM projects are being prepared and conducted in a more disciplined way and their outcomes are continuously monitored and shared through the company's repository. Research limitations/implications To generalize its usefulness, the methodology needs to be applied to the SCM projects of those companies whose organizational and cultural contexts are different from those of Samsung. In addition, the overview of an illustrative SCM project presented in the paper is brief due to space limitations. Practical implications Today, SCM is increasingly recognized as a strategic way to innovate a company's business operation. This paper shows that a methodology such as Samsung's SCM six sigma can be the key to conducting SCM projects in a more disciplined way and for fruitful outcomes. Originality/value The paper introduces a unique six‐sigma‐based methodology for the SCM domain which has been developed and applied in a leading global manufacturing, financial, and services conglomerate. This methodology could be adapted by other companies for their SCM projects to increase the likelihood of project success.

Various deepfake detectors have been proposed, but challenges still exist to detect images of unknown categories or GAN models outside of the training settings. Such issues arise from the overfitting issue, which we discover from our own analysis and the previous studies to originate from the frequency-level artifacts in generated images. We find that ignoring the frequency-level artifacts can improve the detector's generalization across various GAN models, but it can reduce the model's performance for the trained GAN models. Thus, we design a framework to generalize the deepfake detector for both the known and unseen GAN models. Our framework generates the frequency-level perturbation maps to make the generated images indistinguishable from the real images. By updating the deepfake detector along with the training of the perturbation generator, our model is trained to detect the frequency-level artifacts at the initial iterations and consider the image-level irregularities at the last iterations. For experiments, we design new test scenarios varying from the training settings in GAN models, color manipulations, and object categories. Numerous experiments validate the state-of-the-art performance of our deepfake detector.

We study inventory pooling in systems with symmetric costs where supply lead times are endogenously generated by a finite-capacity production system. We investigate the sensitivity of the cost advantage of inventory pooling to various system parameters, including loading, service levels, demand and production time variability, and structure of the production system. The analysis reveals differences in how various parameters affect the cost reduction from pooling and suggests that these differences stem from the manner in which the parameters influence the induced correlation between lead-time demands of the demand streams. We compare these results with those obtained for pure inventory systems, where lead times are exogenous. We also compare inventory pooling with several forms of capacity pooling.

An ever-increasing number of computing devices interconnected through wireless networks encapsulated in the cyber-physical-social systems and a significant amount of sensitive network data transmitted among them have raised security and privacy concerns. Intrusion detection system (IDS) is known as an effective defence mechanism and most recently machine learning (ML) methods are used for its development. However, Internet of Things (IoT) devices often have limited computational resources such as limited energy source, computational power and memory, thus, traditional ML-based IDS that require extensive computational resources are not suitable for running on such devices. This study thus is to design and develop a lightweight ML-based IDS tailored for the resource-constrained devices. Specifically, the study proposes a lightweight ML-based IDS model namely IMPACT (IMPersonation Attack deteCTion using deep auto-encoder and feature abstraction). This is based on deep feature learning with gradient-based linear Support Vector Machine (SVM) to deploy and run on resource-constrained devices by reducing the number of features through feature extraction and selection using a stacked autoencoder (SAE), mutual information (MI) and C4.8 wrapper. The IMPACT is trained on Aegean Wi-Fi Intrusion Dataset (AWID) to detect impersonation attack. Numerical results show that the proposed IMPACT achieved 98.22% accuracy with 97.64% detection rate and 1.20% false alarm rate and outperformed existing state-of-the-art benchmark models. Another key contribution of this study is the investigation of the features in AWID dataset for its usability for further development of IDS.

Despite the active interest in managing organizational knowledge as a strategic resource, most organizations do not yet understand the challenges involved in implementing knowledge management initiatives. Much of the knowledge management literature has been either conceptual or based on individual implementation cases. This study aimed at identifying the several key drivers for developing organizational knowledge management capability and examining their relationships with knowledge management performance. Using data collected from the 66 Korean firms, the study found that knowledge management drivers such as learning orientation, knowledge sharing intention, knowledge management system quality, reward, and knowledge management team activity were significantly related to the organizational knowledge management performance - knowledge quality and user knowledge satisfaction. The study also found that the knowledge management stage of an organization moderates the relationship between some of the knowledge management drivers and knowledge management performance variables.

As widely accepted performance measures in supply chain management practice, frequency-based service levels such as fill rate and stockout rate are often considered in supply contracts under vendor-managed-inventory (VMI) programs. Using a decentralized two-party capacitated supply chain model consisting of one manufacturer and one supplier in a VMI environment, we demonstrate that supplier's service level is in general insufficient for the manufacturer to warrant the desired service level at the customer end. The method by which the supplier achieves her service level to the manufacturer also affects customer service level. By developing bounds on the customer service level, we show that the expected backorders at the supplier should also be taken into account. We suggest a supply contract that offers a menu of different combinations of supplier's service level and expected backorders according to a linear function. Under this contract, the manufacturer can control the end customer service regardless of how the supplier manages her inventory. The supplier has complete flexibility on which combination of the two quantities on the menu to choose according to her own cost functions. Because it does not require any detailed information on supplier's operational characteristics nor her costs, this kind of contract is expected to be easily implementable. In addition, we derive an estimate of the customer service level in terms of the new measures. Our findings have direct implications to supply chain metrics in general: The local service levels are insufficient measures to guarantee the system wide performance. Alternative local measures and/or coordination mechanisms should be employed to achieve desired system performance. Our analysis illustrates a possible way to explore such alternative measures.

The next generation internet will be the internet of things (and not just of computing devices like PCs, PDAs); this is presumed to be enabled by integrating simple computing plus communications capabilities into common objects of everyday use. Radio-frequency identification (RFID) is a compelling technology for creation of such pervasive sensor networks due to its potential for ubiquitous, low-cost/low-maintenance use. However, the current drivers for RFID deployment emphasize supply chain management using passive tags, implying that RFID sensor nets require advances beyond the components and system designs aimed at supply chain applications. This work provides a glimpse of how this may be achieved.

This study aims at providing an alternative view of users' enterprise resource planning (ERP) acceptance. Despite the large body of literature, there are still empirical inquiries to investigate the ERP system implementation from end-users' perspectives as well as from different organizational contexts. To address these issues, we set a project-based sector as our population of interest and seek to understand how project management practices are interrelated with end-users' cognitive perception, and in the end, with their behavioral intention of using the ERP system. In doing so, this study incorporates the best practices of ERP system implementation projects, internal support, external (consultant) support, and functionality selection, into the extended technology acceptance model (TAM) that includes belief constructs and socioenvironmental construct (subjective norm). The empirical analyses show that managerial practices and socioenvironmental factor are significantly related to the original TAM variables in the context of ERP system. One of the interesting findings is the negative effect of consultant support on perceived usefulness, but positive effect on the perceived ease of use, suggesting a useful reference for future research. This study extends the existing literature by investigating potential managerial and socioenvironmental factors affecting user adoption behavior in a different organizational context. This study would also benefit project-based sectors by offering valuable managerial insights that enable them to appreciate and improve end-users' ERP system acceptance and utilization.

This study presents a method for implementing generative AI services by utilizing the Large Language Models (LLM) application architecture. With recent advancements in generative AI technology, LLMs have gained prominence across various domains. In this context, the research addresses the challenge of information scarcity and proposes specific remedies by harnessing LLM capabilities. The investigation delves into strategies for mitigating the issue of inadequate data, offering tailored solutions. The study delves into the efficacy of employing fine-tuning techniques and direct document integration to alleviate data insufficiency. A significant contribution of this work is the development of a Retrieval-Augmented Generation (RAG) model, which tackles the aforementioned challenges. The RAG model is carefully designed to enhance information storage and retrieval processes, ensuring improved content generation. The research elucidates the key phases of the information storage and retrieval methodology underpinned by the RAG model. A comprehensive analysis of these steps is undertaken, emphasizing their significance in addressing the scarcity of data. The study highlights the efficacy of the proposed method, showcasing its applicability through illustrative instances. By implementing the RAG model for information storage and retrieval, the research not only contributes to a deeper comprehension of generative AI technology but also facilitates its practical usability within enterprises utilizing LLMs. This work holds substantial value in advancing the field of generative AI, offering insights into enhancing data-driven content generation and fostering active utilization of LLM-based services within corporate settings.

Achieving high signal-to-noise ratio in chemical and biological sensors enables accurate detection of target analytes. Unfortunately, below the limit of detection (LOD), it becomes difficult to detect the presence of small amounts of analytes and extract useful information via any of the conventional methods. In this work, we examine the possibility of extracting “hidden signals” using deep neural network to enhance gas sensing below the LOD region. As a test case system, we conduct experiments for H2 sensing in six different metallic channels (Au, Cu, Mo, Ni, Pt, Pd) and demonstrate that deep neural network can enhance the sensing capabilities for H2 concentration below the LOD. We demonstrate that this technique could be universally used for different types of sensors and target analytes. Our approach can extract new information from the hidden signals, which can be crucial for next-generation chemical sensing applications and analytical chemistry.

A new performance evaluation paradigm for computer vision systems is proposed. In real situation, the complexity of the input data and/or of the computational procedure can make traditional error propagation methods infeasible. The new approach exploits a resampling technique recently introduced in statistics, the bootstrap. Distributions for the output variables are obtained by perturbing the nuisance properties of the input, i.e., properties with no relevance for the output under ideal conditions. From these bootstrap distributions, the confidence in the adequacy of the assumptions embedded into the computational procedure for the given input is derived. As an example, the new paradigm is applied to the task of edge detection. The performance of several edge detection methods is compared both for synthetic data and real images. The confidence in the output can be used to obtain an edgemap independent of the gradient magnitude.