Vision-and-Language Models (VLMs) have shown impressive capabilities on single-turn benchmarks, yet real-world applications often demand more intricate multi-turn dialogues. Existing multi-turn datasets (e.g, MMDU, ConvBench) only partially capture the breadth and depth of conversational scenarios encountered by users. In this work, we introduce MultiVerse, a novel multi-turn conversation benchmark featuring 647 dialogues - each averaging four turns - derived from a diverse set of 12 popular VLM evaluation benchmarks. With 484 tasks and 484 interaction goals, MultiVerse covers a wide range of topics, from factual knowledge and perception to advanced reasoning tasks such as mathematics and coding. To facilitate robust assessment, we propose a checklist-based evaluation method that leverages GPT-4o as the automated evaluator, measuring performance across 37 key aspects, including perceptual accuracy, linguistic clarity, and factual correctness. We evaluate 18 VLMs on MultiVerse, revealing that even the strongest models (e.g., GPT-4o) achieve only a 50% success rate in complex multi-turn conversations, highlighting the dataset's challenging nature. Notably, we find that providing full dialogue context significantly enhances performance for smaller or weaker models, emphasizing the importance of in-context learning. We believe MultiVerse is a landscape of evaluating multi-turn interaction abilities for VLMs.

Deep learning techniques have emerged as powerful tools for addressing complex and varied problems, achieving remarkable success across numerous AI domains. Despite their effectiveness, the inherent complexity of deep learning models makes them considered black boxes, reducing their interpretability and reliability. To address this challenge, we propose a novel approach called Attribute-guided Relevance Propagation (ARP). ARP enhances the interpretability of deep learning models by learning attributes from specific layers within a pre-trained image classifier and integrating these attributes into saliency maps. This integration not only improves the saliency maps but also identifies and provides example images related to key regions reflected in the maps. We validate the efficacy of ARP through both quantitative and qualitative evaluations, employing widely recognized image classifiers such as ResNet-50 and ViT trained on the benchmark datasets. • Attribute-guided Relevance Propagation (ARP) improves DNN interpretability. • ARP boosts saliency performance for CNNs and Transformers via attribute learning. • ARP provides real examples to support the interpretability of saliency maps. • ARP is validated through extensive experiments on various models and datasets.

Young-Jun Lee, Byungsoo Ko, Han-Gyu Kim, Jonghwan Hyeon, Ho-Jin Choi. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Speech Emotion Recognition (SER) is an important area of research in speech processing that aims to identify and classify emotional states conveyed through speech signals. Recent studies have shown considerable performance in SER by exploiting deep contextualized speech representations from self-supervised learning (SSL) models. However, SSL models pre-trained on clean speech data may not perform well on emotional speech data due to the domain shift problem. To address this problem, this paper proposes a novel approach that simultaneously exploits an SSL model and a domain-agnostic spectral feature (SF) through the Mixture of Experts (MoE) technique. The proposed approach achieves the state-of-the-art performance on weighted accuracy compared to other methods in the IEMOCAP dataset. Moreover, this paper demonstrates the existence of the domain shift problem of SSL models in the SER task.

Object-centric learning (OCL) aspires general and compositional understanding of scenes by representing a scene as a collection of object-centric representations. OCL has also been extended to multi-view image and video datasets to apply various data-driven inductive biases by utilizing geometric or temporal information in the multi-image data. Single-view images carry less information about how to disentangle a given scene than videos or multi-view images do. Hence, owing to the difficulty of applying inductive biases, OCL for single-view images remains challenging, resulting in inconsistent learning of object-centric representation. To this end, we introduce a novel OCL framework for single-view images, SLot Attention via SHepherding (SLASH), which consists of two simple- yet-effective modules on top of Slot Attention. The new modules, Attention Refining Kernel (ARK) and Intermediate Point Predictor and Encoder (IPPE), respectively, prevent slots from being distracted by the background noise and indicate locations for slots to focus on to facilitate learning of object-centric representation. We also propose a weak semi-supervision approach for OCL, whilst our proposed framework can be used without any assistant annotation during the inference. Experiments show that our proposed method enables consistent learning of object-centric representation and achieves strong performance across four datasets. Code is available at https://github.com/object-understanding/SLASH.

In real-world scenarios, detecting out-of-distribution (OOD) action is important when deploying a deep learning-based human action recognition (HAR) model. However, HAR models are easily biased to static information in the video (e.g., background), which can lead to performance degradation of OOD detection methods. In this paper, we propose a simple debiasing framework for out-of-distribution detection in human action recognition. Specifically, our framework eliminates patches with static bias in video using attention maps extracted from the video vision transformer model. Experimental results show that our framework achieves consistent performance improvement on multiple OOD action detection methods and challenging benchmarks. Furthermore, we introduce two new OOD action detection tasks, Kinetics-400 vs. Kinetics-600 exclusive and Kinetics-400 vs. Kinetics-700 exclusive, to validate our method in a setting close to the real-world scenario. With extensive experiments, we demonstrate the effectiveness of our attention-based masking, and in-depth analysis validates the effect of static bias on OOD action detection. The source code and supplementary materials are available at: https://github.com/Simcs/attention-masking.

Anomaly segmentation on the urban landscape scene is an important task in autonomous driving. This process exploits a pre-trained semantic segmentation network to estimate anomalous regions. Anomaly segmentation approaches implemented with extra requirements such as out-of-domain data, extra network, or network retraining might increase the computational cost or degradation of segmentation performance. In this study, to exploit information from the segmentation network for more robust anomaly segmentation, we propose the use of pixel-wise logit variance, which tends to be small for anomalies as network outputs even logits without confidence. Additionally iterative background highlighting is proposed to robustly detect anomalous objects on the background, which is implemented by feeding the logits back into the linear classifier of the network. We achieved state-of-the-art performance among anomaly segmentation approaches without extra requirements, reaching relative average precision improvements of 21.7% on the Fishyscapes Lost&Found and 17.4% on the Fishyscapes Static compared to the state-of-the-art method. The code of this work is available at our Github repository (https://github.com/hagg30/LogitVar).

Reference-based video super-resolution (RefVSR) is a promising domain of super-resolution that recovers high-frequency textures of a video using reference video. The multiple cameras with different focal lengths in mobile devices aid recent works in RefVSR, which aim to super-resolve a low-resolution ultra-wide video by utilizing wide-angle videos. Previous works in RefVSR used all reference frames of a Ref video at each time step for the super-resolution of low-resolution videos. However, computation on higher-resolution images increases the runtime and memory consumption, hence hinders the practical application of RefVSR. To solve this problem, we propose an Efficient Reference-based Video Super-Resolution (ERVSR) that exploits a single reference frame to super-resolve whole low-resolution video frames. We introduce an attention-based feature align module and an aggregation upsampling module that attends LR features using the correlation between the reference and LR frames. The proposed ERVSR achieves 12× faster speed, 1/4 memory consumption than previous state-of-the-art RefVSR networks, and competitive performance on the RealMCVSR dataset while using a single reference image.

Abstract Early-stage fault detection has become an indispensable part of modern industry to prevent potential hazards or sudden hindrances to the production process. With the advent of deep learning (DL) applications in several fields, DL models have been used to classify faults in specific environments. Uniform texture extraction has been performed using transformed-signal processing techniques and deep transfer learning (DTL) architectures in a few studies. Traditional signal processing techniques encounter difficulties in extracting distinct fault features due to the nonlinear and non-stationary nature of the time-series fault data. In this paper, a hybrid DTL architecture comprising a deep convolutional neural network and long short-term memory layers for extracting both temporal and spatial features enhanced by Hilbert transform 2D images is presented. Three standard audio sound fault datasets comprising the malfunctioning industrial machine investigation and inspection dataset, toy anomaly detection in machine operating sounds dataset, and machinery failure prevention technology bearing vibration fault dataset with various loads and noisy environments were utilized in the experimental evaluation. The proposed model with an input size of 32 × 32 achieved an average F 1 score of 0.998 on the tested datasets. The implementation of transfer learning using the three benchmark datasets resulted in the highest accuracy of the proposed model and over fivefold reduction in the training epochs. In addition, the proposed model outperformed the state-of-art models in accuracy in various environments.

Since empathy plays a crucial role in increasing social bonding between people, many studies have designed their own dialogue agents to be empathetic using the well-established method of fine-tuning. However, they do not use prompt-based in-context learning, which has shown powerful performance in various natural language processing (NLP) tasks, for empathetic dialogue generation. Although several studies have investigated few-shot in-context learning for empathetic dialogue generation, an in-depth analysis of the generation of empathetic dialogue with in-context learning remains unclear, especially in GPT-3 (Brown et al., 2020). In this study, we explore whether GPT-3 can generate empathetic dialogues through prompt-based in-context learning in both zero-shot and few-shot settings. To enhance performance, we propose two new in-context example selection methods, called SITSM and EMOSITSM, that utilize emotion and situational information. We also introduce a new automatic evaluation method, DIFF-EPITOME, which reflects the human tendency to express empathy. From the analysis, we reveal that our DIFF-EPITOME is effective in measuring the degree of human empathy. We show that GPT-3 achieves competitive performance with Blender 90M, a state-of-the-art dialogue generative model, on both automatic and human evaluation. Our code is available at https://github.com/passing2961/EmpGPT-3.

In retrieval-based dialogue systems, a response selection model acts as a ranker to select the most appropriate response among several candidates.However, such selection models tend to rely on context-response content similarity, which makes models vulnerable to adversarial responses that are semantically similar but not relevant to the dialogue context.Recent studies have shown that leveraging these adversarial responses as negative training samples is useful for improving the discriminating power of the selection model.Nevertheless, collecting human-written adversarial responses is expensive, and existing synthesizing methods often have limited scalability.To overcome these limitations, this paper proposes a simple but efficient method for generating adversarial negative responses leveraging a large-scale language model.Experimental results on dialogue selection tasks show that our method outperforms other methods of synthesizing adversarial negative responses.These results suggest that our method can be an effective alternative to human annotators in generating adversarial responses.Our dataset and generation code is available at https://github.com/leenw23/ generating-negatives-by-gpt3.Create five irrelevant responses containing keywords of the given dialogue context: ### Dialogue context: """ A: Yes, it's the same every day … B: I'll look forward to that, I can't stand… """ Create five irrelevant responses containing keywords of the given dialogue context: 1. { other examples } ... ...

Detecting emotions from texts is considerably important in an NLP task, but it has the limitation of the scarcity of manually labeled data. To overcome this limitation, many researchers have annotated unlabeled data with certain frequently used annotation procedures. However, most of these studies are focused mainly on English and do not consider the characteristics of the Korean language. In this paper, we present a Korean-specific annotation procedure, which consists of two parts, namely n-gram-based distant supervision and Korean-specific-feature-based distant supervision. We leverage the distant supervision with the n-gram and Korean emotion lexicons. Then, we consider the Korean-specific emotion features. Through experiments, we showed the effectiveness of our procedure by comparing with the KTEA dataset. Additionally, we constructed a large-scale emotion-labeled dataset, Korean Movie Review Emotion (KMRE) Dataset, using our procedure. In order to construct our dataset, we used a large-scale sentiment movie review corpus as the unlabeled dataset. Moreover, we used a Korean emotion lexicon provided by KTEA. We also performed an emotion classification task and a human evaluation on the KMRE dataset.

Abstract With an increase in the number of Web documents, the number of proposed methods for knowledge discovery on Web documents have been increased as well. The documents do not always provide keywords or categories, so unsupervised approaches are desirable, and topic modeling is such an approach for knowledge discovery without using labels. Further, Web documents usually have time information such as publish years, so knowledge patterns over time can be captured by incorporating the time information. The temporal patterns of knowledge can be used to develop useful services such as a graph of research trends, finding similar authors (potential co-authors) to a particular author, or finding top researchers about a specific research domain. In this paper, we propose a new topic model, Author Topic-Flow (ATF) model, whose objective is to capture temporal patterns of research interests of authors over time, where each topic is associated with a research domain. The state-of-the-art model, namely Temporal Author Topic model, has the same objective as ours, where it computes the temporal patterns of authors by combining the patterns of topics. We believe that such ‘indirect’ temporal patterns will be poor than the ‘direct’ temporal patterns of our proposed model. The ATF model allows each author to have a separated variable which models the temporal patterns, so we denote it as ‘direct’ topic flow. The design of the ATF model is based on the hypothesis that ‘direct’ topic flows will be better than the ‘indirect’ topic flows. We prove the hypothesis is true by a structural comparison between the two models and show the effectiveness of the ATF model by empirical results.

In this paper, we improve the low-rank matrix completion algorithm by assuming that the data points lie in a union of low dimensional subspaces. We applied the self-expressiveness, which is a property of a dataset when the data points lie in a union of low dimensional subspaces, to the low-rank matrix completion. By considering self-expressiveness of low dimensional subspaces, the proposed low-rank matrix completion may perform well even with little information, leading to the robust completion on a dataset with high missing rate. In our experiments on movie rating datasets, the proposed model outperforms state-of-the-art matrix completion models. In clustering experiments conducted on MNIST dataset, the result indicates that our method closely recovers the subspaces of original dataset even with the high missing rate.

This paper proposes the rehabilitation treatment coach robot which will help at-home patients do their rehabilitation exercises at home without any professional trainers. The coach robot is designed to be cheap enough for patients to afford it. The robot suggests the rehabilitation program and corrects the posture of the patients during the exercise. The deep neural network is used for posture correction. Besides, the voice interface is applied for convenient interaction between robot and patients during the exercise. The emergency detection module is adopted which will inform doctors when emergency happens on patients. The emergency detection will be implemented using deep neural network on voice input and video input simultaneously. The detailed data collection plan for training deep neural network and performance evaluation plan are also provided in the paper.

Cervical cancer is the women's fourth most common cancer worldwide, with 266,000 deaths in a year. Cervical cancer can be diagnosed by the Papanicolaou test. In this test, a cytopathologist observes a microscopic image of the cervix cells and decides whether the patient is abnormal or not. According to research, the accuracy of the cervical cytology is reported as 89.7%. Because it is associated with the patient's life, it is important to improve the accuracy of this test. Many systems have been proposed to help judge experts to improve the accuracy of tests in the medical field, but development has been limited to areas where there are cleanly quantified test data. In this paper, we design and train a model to automatically classify the normal/abnormal state of cervical cells from microscopic images by using a convolutional neural network and several machine learning classifiers. As a result, the support vector machine achieves the highest performance with 78% F1 score.

The deep neural network (DNN) that models characteristics of general blood test (GBT) results was used in clinical opinions generation. The DNN that generates clinical opinions has the complex structure, which causes overfitting problem. The relatively small size of medical dataset also contributes to the occurrence of overfitting. In order to deal with overfitting, we apply two techniques that solve overfitting of DNN, which are dropout, and batch normalization. Dropout is inserted into the network in various ways in order to find out the optimal structure of the network. Batch normalization is also added in various ways for the same purpose. The experiment conducted on GBT dataset shows that DNNs with dropout and batch normalization outperform the simple DNN in generating clinical opinions for our GBT dataset. Besides, dropout shows slightly better performance compared to batch normalization.

It is still a long way to communicate humans and machines emotionally. There are some tries to provide sentimental conversations among humans and machines. Computational humor is one of research topics in computational linguistics and artificial intelligence. We introduce a new method to generate jokes in a sentence related temporal and spatial contexts for continuous conversations with images. We propose a novel model based on a recurrent neural network with natural language processing (NLP) and understanding (NLU) methods. The method generates jokes in a sentence considering temporal and spatial context. The method can joke to trend sensitive users according to different points of humor that vary from region to region. Through this, the user can feel the interest of the conversational service with humorous responses or contents. We apply the method to some applications such as psychiatric counseling and stress management to enhance the applicability of conversational service.

As the amount of documents continues to increase steadily, it has become an important issue to shorten processing time in the field of natural language processing. In this paper, we describe a method to reduce the execution speed of the Korean temporal information extraction module from a development perspective. While the rule-based approach is useful for finding time representations from natural language sentences, the process of applying rules for each sentence can take a long time. In addition, in Korean sentences, the linguistic characteristics must be considered together to obtain exact temporal information. We first attempted to reduce the time to look for a time expression using the characteristics of the Korean morpheme, and then modified the module to speed up the search for the entire rule base. And we also show an experimental result that the change of execution speed according to the proposed approaches.

There are early studies to attempt users for psychiatric counseling with chatbot. They lead to changes in drinking habit based on intervention approach via chat bot. The application does not consider the user's psychiatric status through the conversations, continuous user monitoring, and ethical judgment in the intervention. We contend that more accurate and continuous emotion recognition gives better satisfaction to users who need mental health care. In addition, appropriate clinical psychological response based on ethical responses is as well. We suggest a conversational service for psychiatric counseling that is adapted methodologies to understand counseling contents based on of high-level natural language understanding (NLU), and emotion recognition based on multi-modal approach. The methodologies enable continuous observation of emotional changes sensitively. In addition, the case-based counseling response model that combines ethical judgment model provides a suitable response to clinical psychiatric counseling.

Dropout is a technique widely used for preventing overfitting while training deep neural networks. However, applying dropout to a neural network typically increases the training time. This paper proposes a different dropout approach called controlled dropout that improves training speed by dropping units in a column-wise or row-wise manner on the matrices. In controlled dropout, a network is trained using compressed matrices of smaller size, which results in notable improvement of training speed. In the experiment on feed-forward neural networks for MNIST data set and convolutional neural networks for CIFAR-10 and SVHN data sets, our proposed method achieves faster training speed than conventional methods both on CPU and GPU, while exhibiting the same regularization performance as conventional dropout. Moreover, the improvement of training speed increases when the number of fully-connected layers increases. As the training process of neural network is an iterative process comprising forward propagation and backpropagation, speed improvement using controlled dropout would provide a significantly decreased training time.

Existing studies in skeleton‐based action recognition mainly utilise skeletal data taken from a single camera. Since the quality of skeletal tracking of a single camera is noisy and unreliable, however, combining data from multiple cameras can improve the tracking quality and hence increase the recognition accuracy. In this study, the authors propose a method called weighted averaging fusion which merges skeletal data of two or more camera views. The method first evaluates the reliability of a set of corresponding joints based on their distances to the centroid, then computes the weighted average of selected joints, that is, each joint is weighted by the overall reliability of the camera reporting the joint. Such obtained, fused skeletal data are used as the input to the action recognition step. Experiments using various frame‐level features and testing schemes show that more than 10% improvement can be achieved in the action recognition accuracy using these fused skeletal data as compared with the single‐view case.

We propose a new customizable tool, Language Independent Feature Extractor (LIFE), which models the inherent patterns of any language and extracts relevant features of thelanguage. There are two contributions of this work: (1) no labeled data is necessary to train LIFE (It works when a sufficient number of unlabeled documents are given), and (2) LIFE is designed to be applicable to any language. We proved the usefulness of LIFE by experimental results of time information extraction.

As documents tend to contain temporal information, extracting such information is attracting much research interests recently.In this paper, we propose a hybrid method that combines machine-learning models and hand-crafted rules for the task of extracting temporal information from unstructured Korean texts.We address Korean-specific research issues and propose a new probabilistic model to generate complementary features.The performance of our approach is demonstrated by experiments on the TempEval-2 dataset, and the Korean TimeBank dataset which we built for this study.

Abstract Natural objects often contain vivid color distribution with wide variety of colors. Conventional colorization techniques, on the other hand, produce colors that are relatively flat with little color variation. In this paper, we introduce a randomized algorithm which considers not only the value of target color but also the distribution of target color. In essence, our algorithm paints a color distribution to a region which synthesizes color distribution of a natural object. Our approach models the correlation between intensity and color in HSV color space in terms of H – S, H – V and S – V joint histogram. During the colorization process, we randomly swap and reassign color of a pixel to minimize a cost function that measures color consistency to its neighborhood and intensity‐to‐color correlation captured in the joint histogram. We tested our algorithm extensively on many natural objects and our user study confirms that our results are more vivid and natural compared to results from previous techniques.

Pervasive healthcare systems, enabled by information and communication technology (ICT), can allow the elderly and chronically ill to stay at home while being constantly monitored. This work aims to present a framework for healthcare monitoring systems based on heterogeneous sensors. In addition to video cameras, the system makes use of a variety of heterogeneous sensors that allow adequate monitoring of each patient by fusing the data from the different sensors. The system described in this paper can be deployed into patients' homes sparing the healthcare costs of living in a health facility where constant monitoring by professionals is required and allowing the elder to live in his home for as long as possible. It can also be deployed in a hospital or a silver-care town environment to cut off the number of healthcare personnel needed, improve the monitoring of patients, and set alarms in case of critical situations.

This paper presents a novel approach to represent human actions in a video. Our approach deals with the limitation of local representation, i.e. space-time interest points, which cannot adequately represent actions in a video due to lack of global information about geometric relationships among interest points. It adds the geometric relationships to interest points by clustering interest points using squared Euclidean distances, followed by using a minimum hexahedron to represent each cluster. Within each video, we build a multi-dimensional histogram based on the characteristics of hexahedrons in the video for recognition. The experimental results show that the proposed representation is powerful to include the global information on top of local interest points and it successfully increases the accuracy of action recognition.

With the development of depth sensors, i.e. Kinect, it is now possible to predict human body poses from a depthmap without any manual labeling. The predicted poses can be used as meaningful features for many applications such as human action recognition. However, existing pose estimation algorithms are not perfect, which can seriously affect the performance of its following applications. In this paper, we propose a novel method to detect erroneous poses. Human poses are captured by Kinect SDK which predicts body joints and connects them with straight lines to represent a pose. We observe depth gradient of pixels located on a body part is consistent when the body part is predicted correctly. With this observation, our algorithm examines depth gradients of pixels on each body part. During the depth gradient processing, our algorithm also considers occlusions. Once a sudden change is detected in depth values on a body part, we check whether the gradient is still consistent excluding the sudden change region. We tested our algorithm on many human activities and our experimental results show that our algorithm acceptably detects erroneous poses in real time.

In addition to search queries and the corresponding clickthrough information, search engine logs record multidimensional information about user search activities, such as search time, location, vertical, and search device. Multidimensional mining of search logs can provide novel insights and useful knowledge for both search engine users and developers. In this paper, we describe our topic-concept cube project, which addresses the business need of supporting multidimensional mining of search logs effectively and efficiently. We answer two challenges. First, search queries and click-through data are well recognized sparse, and thus have to be aggregated properly for effective analysis. Second, there is often a gap between the topic hierarchies in multidimensional aggregate analysis and queries in search logs. To address those challenges, we develop a novel topic-concept model that learns a hierarchy of concepts and topics automatically from search logs. Enabled by the topicconcept model, we construct a topic-concept cube that supports online multidimensional mining of search log data. A distinct feature of our approach is that, in addition to the standard dimensions such as time and location, our topic-concept cube has a dimension of topics and concepts, which substantially facilitates the analysis of log data. To handle a huge amount of log data, we develop distributed algorithms for learning model parameters efficiently. We also devise approaches to computing a topic-concept cube. We report an empirical study verifying the effectiveness and efficiency of our approach on a real data set of 1.96 billion queries and 2.73 billion clicks.

Although synthesis was considered an important and challenging approach to construction of a program or a program model in software development, most of research on synthesis has been devoted to the construction of state machine models or variations of them. Recently, as process modeling through languages like UML Activity Diagram and BPMN appears as a new paradigm of software development, the ability to synthesize models in such languages from requirements would tremendously increase the scope of automatic software development. This paper presents transformation rules for synthesis of UML Activity Diagrams from scenario-based specifications modeled as UML Sequence Diagrams. To that end, we first identify various control flow patterns of Sequence Diagrams and define rules for mapping them to corresponding parts of Activity Diagram. In order to make precise such mapping labeling rules are introduced for the patterns. Also we provide a synthesis algorithm for construction of a UML Activity Diagram from scenarios.

In this paper, we reflect upon the past five years of the KAIST-Carnegie Mellon MSE (Master of Software Engineering) collaboration, and look ahead to the ways in which we can improve in the years to come. With the understanding that the major component of the program lies in its curriculum, our insights focus mainly in the areas of curriculum improvement and evolution. As a means of achieving this goal, two surveys were conducted, one addressing reflections by the program's participating faculty and graduates, and a second investigating various reference curriculums. Based upon the results of both surveys, an improved curriculum structure is proposed, one that identifies and introduces special track options that the authors propose might better serve the needs and demands of Korean industry.

Error concealment is a well-known technique to improve reproduced picture quality in cases which parts of the coded picture are not available at a decoder for reconstruction. The conventional error concealment methods generally make use of the correlation between a damaged marcoblock and its adjacent macroblocks in the same frame and/or the previous frame. However, performance results of concealment methods are different each other and dependent on how much damaged frames and macroblocks are correlated to surrounding ones temporally and spatially. Moreover, most of them can not conceal effectively the consecutively damaged frames or macro blocks caused by wireless bursty packet loss features, since motion vectors and correlation information between the damaged images are already destroyed. The proposed error concealment aware error resilient video coding is a new method which has an intelligent error concealment selector by using pre-estimation technique of the possibly damaged motion vectors at the encoder. The method recommends the best error concealment per macroblock by transmitting additionally its error concealment selection codes to the decoder. The chosen error concealment helps to conceal the damaged MB at the decoder even when losses of frames are bursty. The experimental results show the substantial effectiveness to improve video quality at the cost of a little bit overhead at the transmitted bit streams.

Six Sigma has been adopted by many software development organizations to identify problems in software projects and processes, find optimal solutions for the identified problems, and quantitatively improve the development processes so as to achieve organizations' business goals. A Six Sigma framework for software process improvements is needed to provide a standard process and analysis tools for Six Sigma project executions, and also provide a platform for collaborations with other process improvement approaches, such as PSP/TSP and CMM/CMMI. However, few frameworks have been proposed to support Six Sigma project executions. Most of Six Sigma projects for software process improvements have been performed in an ad-hoc way. In this paper, we propose a framework to support Six Sigma projects for continuous process improvements for software developments. Based on this framework, we implemented a web-based tool, called SSPMT integrated with a software project management tool and a PSP supporting tool. The suggested framework and SSPMT is beneficial in initiating and executing Six Sigma projects, facilitating data collection and data analyses by Six Sigma toolkits, and standardizing the Six Sigma project execution process so as to achieve Six Sigma project goals and of organizations' business goals.

The advent of software process models such as CMM/CMMI (Capability Maturity Model/Capability Maturity Model Integration) has helped software engineers understand principles and approaches of software process improvement. There, however, has been difficulty increasing productivity from applying those models since "how " is not within the scope of the CMM/CMMI. For this reason, SEI (Software Engineering Institute) introduced PSP/TSP (Personal Software Process/Team Software Process); however, they still lack statistical analysis tools and systematic process control techniques for analyzing measures collected in PSP/TSP. Six Sigma, on the other hand, provides the quantitative analysis tools necessary to identify high leverage activities, control process performance and evaluate effectiveness of process changes. Deploying PSP/TSP in conjunction with Six Sigma, therefore, can directly lead to improved project performance and continuous process improvement by analyzing data, assessing process stability, and prioritizing improvements in PSP/TSP. Continuing with this rationale, a framework that guides how and where Six Sigma tools are considered in PSP/TSP is proposed.

Major characteristics of sensor networks, such as node mobility, collaboration among heterogeneous sensors, and environmental changes, require flexible and adaptive software systems. It is also essential to have a mechanism to describe configurations of distributed components and to coordinate the sensors in the network through the configuration information. In this paper, we show how an architectural model can be used to support adaptive software systems for sensor networks. This work is based on our previous work of using an architecture description language (ADL) to represent configurations of software components in sensor nodes and to reconfigure them in response to changing needs of users. We use XML as a means to exchange and manipulate ADL descriptions. To demonstrate how the architecture-based adaptation works for sensor networks, we simulated a scenario of reconfiguring a sensor network to handle node failures.