arXiv subjects

Yunseong Cho

Publications and source records attributed to Yunseong Cho.

12 recordsLinked to original sources

arXiv2026

H-Adapter: Pose-Robust Hairstyle Transfer via Attention-Derived, Source-Aligned Hair Masks

Hairstyle transfer has practical applications such as virtual try-on, yet remains challenging when the source and reference exhibit large head-pose discrepancies. We propose H-Adapter, which improves pose robustness by training with a region-specific loss that disentangles hair and non-hair objectives and thereby induces spatially disentangled cross-attention, from which a source-aligned hair edit mask is derived to guide diffusion-based inpainting. Experiments on pose-agnostic and pose-different subsets demonstrate strong quantitative results, including the best FID, $\mathrm{FID}_{\mathrm{CLIP}}$, and CLIP-I under pose differences, while maintaining competitive non-hair preservation and improving qualitative fidelity to fine-grained reference hairstyle details. Beyond source-conditioned transfer, H-Adapter supports practical extensions including text-to-image generation, auxiliary prompt-based hair color control, and compatibility with an identity-preserving IP-Adapter variant. We also introduce a VLM-as-a-judge protocol and observe consistent gains in hairstyle faithfulness, non-hair preservation, and artifact quality.

cs.CV↗

arXiv2024

Melon Fruit Detection and Quality Assessment Using Generative AI-Based Image Data Augmentation

Monitoring and managing the growth and quality of fruits are very important tasks. To effectively train deep learning models like YOLO for real-time fruit detection, high-quality image datasets are essential. However, such datasets are often lacking in agriculture. Generative AI models can help create high-quality images. In this study, we used MidJourney and Firefly tools to generate images of melon greenhouses and post-harvest fruits through text-to-image, pre-harvest image-to-image, and post-harvest image-to-image methods. We evaluated these AIgenerated images using PSNR and SSIM metrics and tested the detection performance of the YOLOv9 model. We also assessed the net quality of real and generated fruits. Our results showed that generative AI could produce images very similar to real ones, especially for post-harvest fruits. The YOLOv9 model detected the generated images well, and the net quality was also measurable. This shows that generative AI can create realistic images useful for fruit detection and quality assessment, indicating its great potential in agriculture. This study highlights the potential of AI-generated images for data augmentation in melon fruit detection and quality assessment and envisions a positive future for generative AI applications in agriculture.

cs.CV↗

arXiv2024

Near-Field Channel Estimation for XL-RIS Assisted Multi-User XL-MIMO Systems: Hybrid Beamforming Architectures

Channel estimation is one of the key challenges for the deployment of extremely large-scale reconfigurable intelligent surface (XL-RIS) assisted multiple-input multiple-output (MIMO) systems. In this paper, we study the channel estimation problem for XL-RIS assisted multi-user XL-MIMO systems with hybrid beamforming structures. For this system, we propose an {\em unified} channel estimation method that yields a notable estimation accuracy in the near-field BS-RIS and near-field RIS-User channels (in short, near-near field channels), far-near field channels, and far-far field channels. Our key idea is that the effective (or cascaded) channels to be estimated can be each factorized as the product of low-rank matrices (i.e., the product of the common (or user-independent) matrix and the user-specific coefficient matrix). The common matrix whose columns are the basis of the column space of the BS-RIS channel matrix is efficiently estimated via a {\em collaborative} low-rank approximation (CLRA). Leveraging the hybrid beamforming structures, we develop an efficient iterative algorithm that jointly optimizes the user-specific coefficient matrices. Via experiments and complexity analysis, we verify the effectiveness of the proposed channel estimation method (named CLRA-JO) in the aforementioned three classes of wireless channels.

eess.SP↗

arXiv2024

Learning-Based One-Bit Maximum Likelihood Detection for Massive MIMO Systems: Dithering-Aided Adaptive Approach

In this paper, we propose a learning-based detection framework for uplink massive multiple-input and multiple-output (MIMO) systems with one-bit analog-to-digital converters. The learning-based detection only requires counting the occurrences of the quantized outputs of -1 and +1 for estimating a likelihood probability at each antenna. Accordingly, the key advantage of this approach is to perform maximum likelihood detection without explicit channel estimation which has been one of the primary challenges of one-bit quantized systems. However, due to the quasi-deterministic reception in the high signal-to-noise ratio (SNR) regime, one-bit observations in the high SNR regime are biased to either +1 or -1, and thus, the learning requires excessive training to estimate the small likelihood probabilities. To address this drawback, we propose a dither-and-learning technique to estimate likelihood functions from dithered signals. First, we add a dithering signal to artificially decrease the SNR and then infer the likelihood function from the quantized dithered signals by using an SNR estimate derived from a deep neural network-based estimator which is trained offline. We extend our technique by developing an adaptive dither-and-learning method that updates the dithering power according to the patterns observed in the quantized dithered signals. The proposed framework is also applied to channel-coded MIMO systems by computing a bit-wise and user-wise log-likelihood ratio from the refined likelihood probabilities. Simulation results validate the performance of the proposed methods in both uncoded and coded systems.

eess.SP↗

arXiv2023

Coordinated Per-Antenna Power Minimization for Multicell Massive MIMO Systems with Low-Resolution Data Converters

A multicell-coordinated beamforming solution for massive multiple-input multiple-output orthogonal frequency-division multiplexing (OFDM) systems is presented when employing low-resolution data converters and per-antenna level constraints. For a more realistic deployment, we aim to find the downlink (DL) beamformer that minimizes the maximum power on transmit antenna array of each basestation under received signal quality constraints while minimizing per-antenna transmit power. We show that strong duality holds between the primal DL formulation and its manageable Lagrangian dual problem which can be interpreted as the virtual uplink (UL) problem with adjustable noise covariance matrices. For a fixed set of noise covariance matrices, we claim that the virtual UL solution is effectively used to compute the DL beamformer and noise covariance matrices can be subsequently updated with an associated subgradient. Our primary contributions are then (1) formulating the quantized DL OFDM antenna power minimax problem and deriving its associated dual problem, (2) showing strong duality and interpreting the dual as a virtual quantized UL OFDM problem, and (3) developing an iterative minimax algorithm based on the dual problem. Simulations validate the proposed algorithm in terms of the maximum antenna transmit power and peak-to-average-power ratio.

eess.SP↗

arXiv2022

Adaptive Learning-Based Detection for One-Bit Quantized Massive MIMO Systems

We propose an adaptive learning-based framework for uplink massive multiple-input multiple-output (MIMO) systems with one-bit analog-to-digital converters. Learning-based detection does not need to estimate channels, which overcomes a key drawback in one-bit quantized systems. During training, learning-based detection suffers at high signal-to-noise ratio (SNR) because observations will be biased to +1 or -1 which leads to many zero-valued empirical likelihood functions. At low SNR, observations vary frequently in value but the high noise power makes capturing the effect of the channel difficult. To address these drawbacks, we propose an adaptive dithering-and-learning method. During training, received values are mixed with dithering noise whose statistics are known to the base station, and the dithering noise power is updated for each antenna element depending on the observed pattern of the output. We then use the refined probabilities in the one-bit maximum likelihood detection rule. Simulation results validate the detection performance of the proposed method vs. our previous method using fixed dithering noise power as well as zero-forcing and optimal ML detection both of which assume perfect channel knowledge.

eess.SP↗

arXiv2021

Coordinated Beamforming in Quantized Massive MIMO Systems with Per-Antenna Constraints

In this work, we present a solution for coordinated beamforming for large-scale downlink (DL) communication systems with low-resolution data converters when employing a per-antenna power constraint that limits the maximum antenna power to alleviate hardware cost. To this end, we formulate and solve the antenna power minimax problem for the coarsely quantized DL system with target signal-to-interference-plus-noise ratio requirements. We show that the associated Lagrangian dual with uncertain noise covariance matrices achieves zero duality gap and that the dual solution can be used to obtain the primal DL solution. Using strong duality, we propose an iterative algorithm to determine the optimal dual solution, which is used to compute the optimal DL beamformer. We further update the noise covariance matrices using the optimal DL solution with an associated subgradient and perform projection onto the feasible domain. Through simulation, we evaluate the proposed method in maximum antenna power consumption and peak-to-average power ratio which are directly related to hardware efficiency.

eess.SP↗

arXiv2021

Low-Complexity Symbol-Level Precoding for MU-MISO Downlink Systems with QAM Signals

This study proposes the construction of a transmit signal for large-scale antenna systems with cost-effective 1-bit digital-to-analog converters in the downlink. Under quadrature-amplitude-modulation constellations, it is still an open problem to overcome a severe error floor problem caused by its nature property. To this end, we first present a feasibility condition which guarantees that each user's noiseless signal is placed in the desired decision region. For robustness to additive noise, we formulate an optimization problem, we then transform the feasibility conditions to cascaded matrix form. We propose a low-complexity algorithm to generate a 1-bit transmit signal based on the proposed optimization problem formulated as a well-defined mixed-integer-linear-programming. Numerical results validate the superiority of the proposed method in terms of detection performance and computational complexity.

cs.IT↗

arXiv2021

Construction of 1-Bit Transmit Signal Vectors for Downlink MU-MISO Systems: QAM constellations

In this paper, we investigate the construction of a transmit signal for a base station with a massive number of antenna arrays under the cost-effective 1-bit digital-to-analog converters. Due to the coarse nonlinear property, conventional precoding methods could not yield an attractive performance with a severe error-floor problem. Moreover, finding an optimal transmit signal is computationally implausible because of its combinatorial nature. Thus, it is still an open problem to construct a 1-bit transmit signal efficiently. As an extension of our earlier work, we propose an efficient method to construct an 1-bit transmit-signal under quadrature-amplitude-modulation constellations. Toward this, we first derive the so-called feasibility condition which ensures that every user's noiseless observation belongs to a desired decision region, and then transform it as linear constraints. Taking into account the robustness to an additive noise, the proposed construction method is formulated as a well-defined mixed-integer-linear-programming problem. Based on this, we develop a low-complexity algorithm to solve it (equivalently, to generate a 1-bit transmit signal). Via simulations, we verify the superiority of the proposed method in terms of a computational complexity and detection performance.

cs.IT↗

arXiv2021

Coordinated Multicell Beamforming and Power Allocation for Massive MIMO with Low-Resolution ADC/DAC

In this work, we present a solution for coordinated beamforming and power allocation when base stations employ a massive number of antennas equipped with low-resolution analog-to-digital and digital-to-analog converters. We address total power minimization problems of the coarsely quantized uplink (UL) and downlink (DL) communication systems with target signal-to-interference-plus-noise ratio (SINR) constraints. By combining the UL problem with minimum mean square error combiners and deriving the Lagrangian dual of the DL problem, we prove UL-DL duality and show there is no duality gap even with coarse data converters. Inspired by strong duality, we devise an iterative algorithm to determine the optimal UL transmit powers, and then linearly amplify the UL combiners with proper weights to acquire the optimal DL precoder. Simulation results validate strong duality and evaluate the proposed method in terms of total power consumption and achieved SINR.

cs.IT↗

arXiv2020

Quantized Massive MIMO Systems with Multicell Coordinated Beamforming and Power Control

In this paper, we investigate a coordinated multipoint (CoMP) beamforming and power control problem for base stations (BSs) with a massive number of antenna arrays under coarse quantization at low-resolution analog-to-digital converters (ADCs) and digital-to-analog converter (DACs). Unlike high-resolution ADC and DAC systems, non-negligible quantization noise that needs to be considered in CoMP design makes the problem more challenging. We first formulate total power minimization problems of both uplink (UL) and downlink (DL) systems subject to signal-to-interference-and-noise ratio (SINR) constraints. We then derive strong duality for the UL and DL problems under coarse quantization systems. Leveraging the duality, we propose a framework that is directed toward a twofold aim: to discover the optimal transmit powers in UL by developing iterative algorithm in a distributed manner and to obtain the optimal precoder in DL as a scaled instance of UL combiner. Under homogeneous transmit power and SINR constraints per cell, we further derive a deterministic solution for the UL CoMP problem by analyzing the lower bound of the SINR. Lastly, we extend the derived result to wideband orthogonal frequency-division multiplexing systems to optimize transmit power and beamformer for all subcarriers. Simulation results validate the theoretical results and proposed algorithms.

cs.IT↗

arXiv2019

Robust Learning-Based ML Detection for Massive MIMO Systems with One-Bit Quantized Signals

In this paper, we investigate learning-based maximum likelihood (ML) detection for uplink massive multiple-input and multiple-output (MIMO) systems with one-bit analog-to-digital converters (ADCs). To overcome the significant dependency of learning-based detection on the training length, we propose two one-bit ML detection methods: a biased-learning method and a dithering-and-learning method. The biased-learning method keeps likelihood functions with zero probability from wiping out the obtained information through learning, thereby providing more robust detection performance. Extending the biased method to a system with knowledge of the received signal-to-noise ratio, the dithering-and-learning method estimates more likelihood functions by adding dithering noise to the quantizer input. The proposed methods are further improved by adopting the post likelihood function update, which exploits correctly decoded data symbols as training pilot symbols. The proposed methods avoid the need for channel estimation. Simulation results validate the detection performance of the proposed methods in symbol error rate.

cs.IT↗