To create culturally inclusive vision-language models (VLMs), developing a benchmark that tests their ability to address culturally relevant questions is essential. Existing approaches typically rely on human annotators, making the process labor-intensive and creating a cognitive burden in generating diverse questions. To address this, we propose a semi-automated framework for constructing cultural VLM benchmarks, specifically targeting multiple-choice QA. This framework combines human-VLM collaboration, where VLMs generate questions based on guidelines, a small set of annotated examples, and relevant knowledge, followed by a verification process by native speakers. We demonstrate the effectiveness of this framework through the creation of K-Viscuit, a dataset focused on Korean culture. Our experiments on this dataset reveal that open-source models lag behind proprietary ones in understanding Korean culture, highlighting key areas for improvement. We also present a series of further analyses, including human evaluation, augmenting VLMs with external knowledge, and the evaluation beyond multiple-choice QA. Our dataset is available at https://huggingface.co/datasets/ddehun/k-viscuit.

With the ability to collect vast amounts of image and natural language data from the web, there has been a remarkable advancement in Large-scale Language Models (LLMs). This progress has led to the emergence of chatbots and dialogue systems capable of fluent conversations with humans. As the variety of devices enabling interactions between humans and agents expands, and the performance of text-based dialogue systems improves, there has been recently proposed research on visual dialog. However, visual dialog requires understanding sequences of pairs consisting of images and sentences, making it challenging to gather sufficient data for training large-scale models from the web. In this paper, we propose a new multimodal learning method leveraging existing large-scale models designed for each modality, to enable model training for visual dialog with small visual dialog datasets. The key ideas of our approach are: 1) storing the history or context during the progression of visual dialog in the form of spatiotemporal graphs, and 2) introducing small modulation blocks between modality-specific models and the graphs to align the semantic spaces. For implementation, we introduce a novel structure-aware cross-attention method, which retrieves relevant image and text knowledge for utterance generation from the pretrained models. For experiments, we achieved a new state-of-the-art performance on three visual dialog datasets, including the most challenging one COMET.

Movement is one of the basic tools that humans use to convey emotional states. Body language and movement are also relevant in the perception that we have of other people. Many studies have been done on movement and emotion sharing, involving humans on one side and robots or automated agents on the other. From these studies, there is evidence of the importance of robots to improve their social capabilities and develop more effective social interaction. This work aims at embedding some social movements in a robot manipulator, that will elicit an emotional response from users. Laban Movement Analysis is used to do so, and social movements are developed on a robotic arm, then shown to participants in an online study and a questionnaire. The results show that it is possible to elicit emotions through movement only and that the perception is not affected by personal experiences.

In recent years, autonomous robots have proven capable of solving tasks in complex environments. In particular, robot manipulations in activities of daily living (ADL) for service robots have been widely developed. However, manipulations of grasping an unreachable object in domestic environments still present difficulty. To perform those applications better, we developed an autonomous system with human-in-the-loop that combined the cognitive skills of a human operator with autonomous robot behaviors. In this work, we present techniques for integration the system for assistive mobile manipulation and new strategies to support users in the domestic environment. We demonstrate that the robot can grasp multiple objects with random size at known and unknown table heights. Specifically, we developed three strategies for manipulation. We also demonstrated these strategies using two intuitive interfaces, a visual interface in rviz and a voice user interface with speech recognition. Moreover, the robot can select strategies automatically in random scenarios, which make the robot intelligent and able to make decisions independently in the environment. We demonstrated that our robot shows the capabilities for employment in domestic environments to perform actual tasks.

Millimeter wave MIMO systems provide orders of magnitude in data rates due to larger bandwidth than conventional MIMO systems. On the other hand, they experience severe free-space pathloss as carrier frequency ten-fold increases. The resulting small wavelength helps to overcome pathloss with beamforming gain using massive antennas, known as precoding. In mmWave MIMO system, hybrid beamforming combines analog beamforming with digital precoding to reduce the energy consumption and complexity. Hybrid beamforming uses a smaller number of RF chains, which have large power consumption, than the number of antennas. In order to design analog and digital precoder, a large number of calculations such as the SVD of channel is required. In this paper, we propose low complexity hybrid precoding algorithm using beam steering, which utilizes array response vectors of the channel. We first propose the method of selection the analog beamforming matrix without any SVD or inversion of matrices. Then, we find the digital precoding matrix which performs a maximum of the total sum rate with fixed analog beamforming matrix. Since our proposed algorithm applies a set of array response vectors that are used to form the channel, there is no need for complicated operations as SVD like traditional precoding algorithms. Simulation results show that proposed algorithm can reduce the computational complexity while achieving about 93% the same performance in achievable rate compared to typical hybrid precoding in mmWave MIMO systems.

In this paper, we propose a minimum-mean-squared-error (MMSE)-based lattice-reduction (LR)-aided fixed-complexity sphere decoder (FSD) for low-complexity near-maximum-likelihood (near-ML) multiple-input multiple-output (MIMO) detection. In order for the FSD to achieve optimal performance, the number of full expansion (FE) stages should be sufficient, which is the major cause of the increase in the computational complexity when either a large signal constellation or a large number of antennas are adopted. However, the proposed algorithm maintains the near-ML performance with the aid of the MMSE-based LR algorithm while reducing the number of FE stages. Although there exists the increase in the computational complexity for the application of the additional processing elements, the decrease in the number of FE stages results in the lower computational complexity of the overall algorithm. The numerical analysis demonstrates that there is a considerable decrease in the computational complexity while the performance degradation is negligible, compared to the optimal FSD.

In this letter, we propose an exact maximum likelihood (ML) criterion based on semidefinite relaxation (SDR) in multiple-input multiple-output systems. Although a conventional SDR criterion for determining whether a symbol is the ML solution exists, its results cannot be guaranteed when noise is present. In place of the conventional criterion's positive semidefinite (PSD) discriminant, we propose a new, exact ML criterion based on the condition that all diagonal values are positive (PDV), a simple characteristic and necessary condition of PSD. The proposed criterion has a lower calculation complexity for testing than does a PSD and can ensure that the ML solution is always satisfactory.

In this paper, we propose a high performance multiple-input and multiple-output (MIMO) detection algorithm for downlink multiuser MIMO (DL MU-MIMO) with practical errors in IEEE 802.11ac systems. The DL MU-MIMO system using the precoding matrix, which is generated through imperfect channel state information caused by channel estimation error and channel feedback quantization error, has severe performance degradation since each station (STA) receives the desired signal for the STA as well as the interference signal for different STAs. Therefore, the proposed detection algorithm is developed by considering both the desired signal and the interference signal. Using the characteristics of the interference signal, the proposed detection algorithm is performed by additionally detecting several interference symbols. The proposed algorithm has better performance than the conventional algorithm considering only the desired signal. Simulation results show that the coded bit error rate performance of the proposed algorithm was improved in all modulation and coding scheme cases, compared to the conventional algorithm.

This paper proposes an efficient hybrid modulation scheme for a mixed phase-shift keying and pulse-position modulation (PSK/PPM) for heterogeneous networks employing ultra-wideband impulse radio (UWB-IR). The proposed scheme can support both coherent and noncoherent reception and is compatible with the IEEE 802.15.4a standard. Conventional hybrid modulation schemes for heterogeneous networks are generally constructed with a super-orthogonal coding (SOC) technique with which coherent receivers can get an additional performance improvement by encoding the phase bits. However, with the proposed modulation scheme, a coherent receiver can achieve a performance improvement by displaying its best capability on phase detection through a simple waveform mapping scheme. The proposed scheme requires no extra hardware or power consumption for performance improvement in coherent receivers. Simulations show that the BER of the proposed scheme in a coherent receiver is increased by 2.7dB over the hybrid modulation with no combined phase coding and also comparable to the SOC-combined modulation scheme employed in the IEEEE 802.15.4a. The error performance of the proposed scheme is analyzed, and the numerical results are presented in this paper.

In this paper, we propose an antenna selection (AS) scheme with reduced feedback for enhanced WiMedia ultra-wideband (UWB) systems with multiple transmit and receive antennas. The proposed scheme groups several adjacent subcarriers and selects one transmit antennas per group. Only indexes for the selected antennas are transmitted to reduce the amount of feedback information. We also present a practical operating scenario for the proposed AS scheme. By following the scenario, the proposed scheme not only can prevent throughput loss due to feedback transmission but can also provide robustness against channel changes between several frames. From simulation results for 1024 Mbps transmission mode in the WiMedia UWB system, the proposed scheme showed that performance was improved by 7 dB and transmission range was extended to 4.7 m at 1% PER, compared to the orignal 1 × 1 WiMedia UWB standard.

This paper presents a new approach for reducing transmitter-leakage in UHF RFID readers for mobile devices. In the approach, it is possible to reduce the leakage dynamically by utilizing pilot signals for detecting the leakage and then applying reverse-phase signals for its compensation. Experimental results show more than 30dB of the leakage elimination and 25∼600% of enhancement in the reading range when tested with a 8dBi stationary and a −3.5dBi mobile antennae in an anechoic chamber.

This paper proposes an edge-directed interpolation algorithm to enhance the quality of natural images which are captured by low-resolution camera installed on car or CCTV. Based on the accurate estimation of edge directional covariance between low-resolution and high-resolution image, local covariance coefficients extracted from the low-resolution image has been adapted for the interpolation to obtain the high-resolution image. DCT (Discrete Cosine Transform) kernel function is used in order to reflect the multi-directional edge accurately without increasing of complexity. Simulation result shows that our new interpolation algorithm significantly improves the subjective quality of the interpolated images compared with conventional linear interpolation and NEDI one. It also demonstrates the improvements of objective metrics such as PSNR, SSIM(structural similarity index measurement) and WEA (Wiener filter coefficients Estimation Accuracy) which are used for the accuracy estimation of directionality.

NA

In this paper, we propose an efficient design and implementation results of a high speed 2TX-2RX multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) wireless LAN (WLAN) baseband processor. The proposed processor includes bit-parallel processing transmitter physical layer convergence procedure (TX-PLCP) processor and space-division multiplexing (SDM) symbol detector, which have been optimized for low power consumption and low hardware overhead. It was implemented using 0.18-µm CMOS technology. The proposed architecture can operate at a 40-MHz clock frequency and supports the maximum data rate of 130Mbps. The logic gate count for the processor is 978K and the power consumption is 62 / 284mW (TX / RX), respectively.

In this paper we suggested a new image watermarking algorithm which can hide special information of the ID owner for preventing ID's photo forgery. The algorithm should have robustness against to printing and scanning of the photo because the photo used in ID isn't a digital photo but a printed analog photo. And it also should have enough payloads to embed an identical number like a social security number into a small photo. The robustness and the sufficient payloads are guaranteed by the proposed peak position modulation of spread spectrum correlation. As a result we can embed and extract 40 bits information at 256 × 256 size photo with BER = 0 when the test condition is 300 dpi scanner and photo printer, and 22 ID photos were used. In addition, compared with other image watermarking techniques, the algorithm could reduce the computational cost because a transformation or equipment is not required.

This paper proposes a new I/Q imbalance estimation algorithm for multiband orthogonal frequency division multiplexing (MB-OFDM) based ultra-wideband (UWB) systems. The proposed algorithm is highly effective for the UWB channels with strong frequency selectivity in the presence of carrier frequency offset (CFO). In order to estimate I/Q imbalance, the 2nd-order moments of received signal are analyzed, and a metric function is derived considering CFO. The I/Q imbalance is estimated based on this function which can be sequentially calculated with only two preamble symbols per each sub-band in timedomain, which ensures immediate convergence and robustness to frequency selectivity. The performance of the algorithm is evaluated by simulations in the case of 480 Mbps transmission mode in WiMedia MB-OFDM system. The simulation results show that the proposed algorithm achieves about 4 dB SNR improvement over the existing methods at BER=10-5.

In this paper, we propose a low complexity interference detection scheme for the DAA technique in the WiMedia UWB system. According to definition of DAA in worldwide radio frequency, many WiMAX signal detection schemes have been studied and time-domain coherent detection (TDCD) scheme is the best approach with respect to the detection performance. However, this scheme requires the additional filter and down-sampler to reconstruct WiMAX signal from the received signal and these requirements lead to increase of hardware complexity. In order to reduce the hardware complexity without performance degradation, the proposed scheme is based on cross-correlation in frequency-domain. Simulation results show that the detection performance of the proposed scheme is almost the same as that of TDCD scheme and computational complexity is about 50% less than that of the TDCD scheme.

The thrust of this paper is to profile the execution phases of applications, which helps optimize the efficiency of the underlying resources. Here we present a novel system-level application-resource-demand phase analysis and prediction approach in support of on-demand resource provisioning. The process we follow is to explore large-scale behavior of applications’ resource consumption, followed by analysis using a set of algorithms based on clustering. The phase profile, which learns from historical runs, is used to classify and predict future phase behavior. This process takes into consideration applications’s resource consumption patterns, phase transition costs and penalties associated with Service-Level Agreements (SLA) violations. Our experimental results with WorldCup98 replay web access logs show that prediction accuracies around 84% or larger for ten-phase cases can be achieved for network performance traces.

In this letter, we propose two efficient symbol detection algorithms for OFDM-based wireless LANs with space-frequency transmit diversity (SF-TD) scheme. In case that SF-TD scheme is applied to OFDM-based wireless LANs, the interference caused by insufficient number of sub-carriers may be generated between adjacent sub-carriers, and it degrades the overall system performance seriously. The proposed algorithms eliminate this interference in a parallel or sequential manner and achieve a considerable performance improvement over a conventional detection algorithm

The rate control is important to allocate bits efficiently for getting better performance, such as high quality, low fluctuation of quality and low mismatch between a target bit rate and an encoded bit rate. In this paper, we present an improved frame-layer H.264/AVC rate control scheme using enhanced complexity measure, scene-change detection, and quantization parameter adjustment for low bit rate video. Especially, our complexity measure contains not only the residual information but also the motion information. We use the motion vector difference encoded previously as motion information for low computation-load encoder. Experimental results show that Y-PSNR and bit-rate mismatch are remarkably improved and the fluctuation of Y-PSNR is similar as compared with the conventional H.264/AVC rate control scheme (JVT-G012)

This letter proposes a new distributed arithmetic (DA) algorithm for low-power finite-impulse response (FIR) filter implementation. The characteristic of the proposed algorithm is that the FIR filters using the proposed algorithm do not need to employ two's complement representation in lookup tables as well as multiply-and-accumulation blocks. Thus, the proposed algorithm can minimize the dynamic power consumption of the FIR filters. The experimental results show that the lowpass FIR filter using the proposed algorithm achieves 29% and 26% power consumption reduction compared to that using the conventional algorithm for zero-mean random inputs and speech inputs, respectively.

We propose a color interpolation method using high order B-splines for a single-chip CCD (charge-coupled device). Single-chip digital cameras use a color filter array (CFA) and an interpolation method in order to reconstruct high quality images from sparsely sampled images. Frequently, single-chip digital cameras use the bilinear interpolation method due to the simple implementation and fast processing time. We have applied the interpolation algorithm using the cubic spline in order to obtain high quality images. Experiments show that the cubic spline interpolation method provides better performance than the bilinear interpolation method.

In this paper, the problem of optimal PN sequence design for quasi-synchronous (QS) DS-CDMA communication systems is formulated, and classes of optimal PN sequences which have good pseudorandomness and small cross-correlation values for some range of offsets are designed. The new results are compared with the commonly used PN sequences, and the system performances of the QS CDMA systems using different PN sequences are also evaluated.

We &veloped a new meth~called MP2, for partitioning networks into multiple (> 2) blocks each of which has both size and pin constraints.The MP2 method uses an improvement approach snd tries to minimize the total number of terrnintds of tdl blocks while sstistjing the pin and size constraints of every block It supports multiple classes of cells in input networks and blocks.It makes use of a scalar value of benejlt which captures lookshetd information.It is the tit improvement method tbst considers pin constraints of blocks.It has been applied to partitioning tedtrtology-mspped circuits into multiple FFGA chips.In sddhion to describing the MP2 meth~we will discuss some interesting iindings we gleaned during our experiments.