JCJeihee Cho

Jeihee Cho

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Jeihee Cho has a total of 12 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Blind-QEM: A Privacy-Preserving Framework for Quantum Error Mitigation via Zero-Knowledge Proofs

    by Junyong Lee, Jeihee Cho, Hyeonseong Jung, Euimin Lee, Yunah Choi, Seokhyeon Son, Shiho Kim

    As quantum computing moves to a cloud-based service model, a privacy–utility dilemma arises: effective Quantum Error Mitigation (QEM) requires circuit visibility, yet circuits and noise models are often proprietary. We propose Blind-QEM, a privacy-preserving framework that enables outsourced mitigation without revealing circuit topology. Using Zero-Knowledge Proofs (ZKPs) and a receipt-based binding mechanism anchored by QPU-signed execution logs, Blind-QEM verifies policy compliance and cryptographically links results to committed circuits. This allows Service Providers to perform global incoherent noise cancellation and readout mitigation using only verified aggregate statistics, ensuring mutual protection of user IP and SP models.

  • Open Access

    Trustless Feedback: Preserving Privacy and Integrity in Crowdsourced RLHF via Zero-Knowledge Proofs

    by Jihoon Oh, Junyong Lee, Jeihee Cho, Hyeonseong Jung, Eumin Lee, Yunah Choi, Shiho Kim


  • Open Access

    MaaI: ML-as-an-Infrastructure for Diverse Personalized Inference With Edge Offloading

    by Junyong Lee, Jeihee Cho, Shiho Kim

    The rapid advancements in large language models (LLMs) have transformed applications across industries, from natural language understanding to content generation. However, their escalating computational demands and dependence on cloud services create barriers to the widespread adoption of customized models. We introduce MaaI, an edge-offloading architecture that optimizes LLM deployment by distributing computational tasks and customization across edge devices and cloud servers. Crucially, MaaI identifies the optimal partition point to divide the model between the server and edge, ensuring that foundational components are allocated to edge devices while intensive computations remain in the cloud. Utilizing quantization and LoRA adapters, this approach enables users to leverage larger models with enhanced performance while minimizing server load. Furthermore, by retaining only the foundational model on the server and offloading customized adaptations to edge devices, MaaI facilitates scalable, personalized deployments without significantly increasing server resource requirements. Experimental results across diverse benchmarks and hardware configurations confirm MaaI’s effectiveness in optimizing performance, memory efficiency, and latency.

  • Predictive Path Planning of Multiple UAVs for Effective Network Hotspot Coverage

    by JeiHee Cho, SooMin Ki, HyungJune Lee

    In event or disaster scenarios where network communication is jammed, it is important to provide stable network service to users within a reasonable amount of time. We propose a path planning algorithm for unmanned aerial vehicles (UAVs) to serve network traffic in hotspot areas using spatio-temporal information about requests among the region of interest (RoI). The main task of a UAV is to provide communication services to users, while preparing for future hotspots. We propose a simple yet efficient trajectory design consisting of two phases: 1) targeting traffic for a single UAV, and 2) cooperative targeting for multiple UAVs. First, each UAV selects a long-term target considering future traffic and then a short-term target considering the present traffic. When UAVs encounter other UAVs, a cooperative targeting phase ensures UAVs serve traffic in different locations or with different statuses. Our trajectory design enables a UAV to construct its own path for a continuous UAV-enabled network. Simulation and real-world dataset-based experiments confirmed that our targeting scheme provides sufficient network service in a reasonable time, with an average service rate factor of up to 0.85, and an average service completion time relative to the deadline of up to 0.23. The experimental results have demonstrated that our proposed algorithm provides more stable performance compared to other existing algorithms.


  • Open Access

    Towards Persistent Surveillance and Reconnaissance Using a Connected Swarm of Multiple UAVs

    by Jeihee Cho, Jaeyi Sung, Jinyi Yoon, Hyungjune Lee

    In situations where surveillance or communication infrastructure has collapsed, it is important to keep monitoring affected areas. We leverage unmanned aerial vehicles (UAVs) to collect and provide up-to-date on-site information to a data consumer in an efficient way, for later complete yet agile analysis. We propose a distributed dynamic data collection scheme for persistent surveillance and reconnaissance, using a swarm of connected UAVs with two phases of operation: 1) network formation; and 2) UAV traversal of a region of interest. The main task of a UAV is to continuously collect data within its sensing range, while the UAV swarm travels along the calculated paths. When UAVs are newly connected to form a swarm, or disconnected from an already-formed swarm, a formation phase begins. In the formation phase, UAVs become a single group and produce a compact, dynamically alternating formation called DiagonalX to cover broad areas, including boundary parts, in a fair and effective manner. During the traversal phase, each UAV swarm finds a simple yet efficient navigation path based on data freshness to cover sub-areas and continuously obtain up-to-date information evenly throughout the whole region of interest. Simulation experiments confirm that both formation and traversal procedures perform essential tasks in a distributed manner, while maintaining better data freshness than other counterpart algorithms, with a freshness factor of up to 5.77, and reasonable overheads. An additional feature, a dynamically aperiodic formation change, achieves a more stable performance.