DSpace Community: KAIST School of Computing
http://hdl.handle.net/10203/8
KAIST School of Computing2024-03-13T17:14:54ZGPUTucker: Large-Scale GPU-Based Tucker Decomposition Using Tensor Partitioning
http://hdl.handle.net/10203/314348
Title: GPUTucker: Large-Scale GPU-Based Tucker Decomposition Using Tensor Partitioning
Authors: Lee, Jihye; Han, Donghyoung; Kwon, Oh-Kyoung; Chon, Kang-Wook; Kim, Min-Soo
Abstract: Tucker decomposition is used extensively for modeling multi-dimensional data represented as tensors. Owing to the increasing magnitude of nonzero values in real-world tensors, a growing demand has emerged for expeditious and scalable Tucker decomposition techniques. Several graphics processing unit (GPU)-accelerated techniques have been proposed for Tucker decomposition to decrease the decomposition speed. However, these approaches often encounter difficulties in handling extensive tensors owing to their huge memory demands, which exceed the available capacity of GPU memory. This study presents an expandable GPU-based technique for Tucker decomposition called GPUTucker. The proposed method meticulously partitions sizable tensors into smaller sub-tensors, which are referred to as tensor blocks, and effectively implements the GPU-based data pipeline by handling these tensor blocks asynchronously. Extensive experiments demonstrate that GPUTucker outperforms state-of-the-art Tucker decomposition methods in terms of the decomposition speed and scalability.2024-03-01T00:00:00ZSuM: Efficient shadow stack protection on ARM Cortex-M
http://hdl.handle.net/10203/314778
Title: SuM: Efficient shadow stack protection on ARM Cortex-M
Authors: Choi, Wonwoo; Seo, Minjae; Lee, Seongman; Kang, Brent Byunghoon
Abstract: System software written in unsafe languages such as C/C++ is susceptible to various types of security vulnerabilities. Historically, backward-edges such as return addresses have been an attractive target for control-flow hijacking attacks due to the severity and ease of exploitation. Although various backward-edge control-flow integrity schemes have been proposed over the years, most of them mainly focus on protecting desktop/server-class systems, leaving embedded systems unprotected. Even worse, bringing their defense mechanisms into resource-constrained embedded systems is undesirable because they were originally designed for high-end computing systems and thus are not directly applicable to embedded systems without compromising performance and real-time constraints. In this paper, we propose Shadow under the Mask (SUM), an efficient and robust backward-edge control flow protection that is applicable to ARM Cortex-M processors. Specifically, SUM realizes a non-bypassable shadow stack mechanism and safeguards its structural integrity in a novel combination of an MPU and FaultMask—an overlooked hardware feature in Cortex-M processors. To be more specific, SUM restricts all access to the shadow stack through MPU, ensuring its integrity; and temporarily disables its MPU protection through FaultMask during the execution of safe instructions, guaranteeing that only authorized instructions can modify the shadow stack. In our empirical evaluation, SUM incurs minimal runtime overhead of 2.77% and 2.63%, respectively, on the BEEBS and CoreMark benchmark suites. These results underscore the viability of our proposed approach as a practical and potent solution to address the highlighted cybersecurity challenge.2024-01-01T00:00:00ZEC: A Tool for Guiding Chart and Caption Emphasis
http://hdl.handle.net/10203/317282
Title: EC: A Tool for Guiding Chart and Caption Emphasis
Authors: Kim, Dae Hyun; Choi, Seulgi; Kim, Juho; Setlur, Vidya; Agrawala, Maneesh
Abstract: Recent work has shown that when both the chart and caption emphasize the same aspects of the data, readers tend to remember the doubly-emphasized features as takeaways; when there is a mismatch, readers rely on the chart to form takeaways and can miss information in the caption text. Through a survey of 280 chart-caption pairs in real-world sources (e.g., news media, poll reports, government reports, academic articles, and Tableau Public), we find that captions often do not emphasize the same information in practice, which could limit how effectively readers take away the authors' intended messages. Motivated by the survey findings, we present EMPHASISCHECKER, an interactive tool that highlights visually prominent chart features as well as the features emphasized by the caption text along with any mismatches in the emphasis. The tool implements a time-series prominent feature detector based on the Ramer-Douglas-Peucker algorithm and a text reference extractor that identifies time references and data descriptions in the caption and matches them with chart data. This information enables authors to compare features emphasized by these two modalities, quickly see mismatches, and make necessary revisions. A user study confirms that our tool is both useful and easy to use when authoring charts and captions.2024-01-01T00:00:00ZInverse Constraint Learning and Generalization by Transferable Reward Decomposition
http://hdl.handle.net/10203/315435
Title: Inverse Constraint Learning and Generalization by Transferable Reward Decomposition
Authors: Jang, Jaehwi; Song, Minjae; Park, Daehyung
Abstract: We present the problem of inverse constraint learning (ICL), which recovers constraints from demonstrations to autonomously reproduce constrained skills in new scenarios. However, ICL suffers from an ill-posed nature, leading to inaccurate inference of constraints from demonstrations. To figure it out, we introduce a transferable constraint learning (TCL) algorithm that jointly infers a task-oriented reward and a task-agnostic constraint, enabling the generalization of learned skills. Our method TCL additively decomposes the overall reward into a task reward and its residual as soft constraints, maximizing policy divergence between task- and constraint-oriented policies to obtain a transferable constraint. Evaluating our method and five baselines in three simulated environments, we show TCL outperforms state-of-the-art IRL and ICL algorithms, achieving up to a 72% higher task-success rates with accurate decomposition compared to the next best approach in novel scenarios. Further, we demonstrate the robustness of TCL on two real-world robotic tasks.2024-01-01T00:00:00Z