Random chunks generation attack is a more sophisticated form of side channel attack in the process of cross-user client side deduplication. By constructing a deduplication request consists of one or more target chunks together with a number of randomly generated non-duplicate ones, the situation that every one of the target chunks is duplicate cannot be differentiated from that none or part of them is duplicate, which hinders the introduction of obfuscation. Once such an attack is launched in a hybrid mode, the risk is even more significant since a proportion of target chunks is replaced with public ones, which makes the existence privacy of remaining target chunks easier to be exposed. In order to deal with these challenges, we propose a novel grouping strategy based secure cross-user deduplication scheme for cloud storage, which takes the lead to achieve security against the hybrid attack mentioned above. Specifically, we propose a secure grouping obfuscation strategy and an aggregated response generation mechanism, allowing the cloud service provider to randomly distribute query tags of chunks in a certain request into several groups, and generate an obfuscated response for each one of them according to an adaptive response table before returning an aggregated result. The security analysis and experimental results show that the proposed scheme is able to achieve resistance against the aforementioned attacks with controllable overhead introduced no matter how many target chunks are involved.

Computation-intensive tensor operators constitute over 90% of the computations in Large Language Models (LLMs) and Deep Neural Networks. Automatically and efficiently generating high-performance tensor operators with hardware primitives is crucial for diverse and ever-evolving hardware architectures like RISC-V, ARM, and GPUs, as manually optimized implementation takes at least months and lacks portability. LLMs excel at generating high-level language codes, but they struggle to fully comprehend hardware characteristics and produce high-performance tensor operators. We introduce a tensor-operator auto-generation framework with a one-line user prompt (QiMeng-TensorOp), which enables LLMs to automatically exploit hardware characteristics to generate tensor operators with hardware primitives, and tune parameters for optimal performance across diverse hardware. Experimental results on various hardware platforms, SOTA LLMs, and typical tensor operators demonstrate that QiMeng-TensorOp effectively unleashes the computing capability of various hardware platforms, and automatically generates tensor operators of superior performance. Compared with vanilla LLMs, QiMeng-TensorOp achieves up to 1291× performance improvement. Even compared with human experts, QiMeng-TensorOp could reach 251% of OpenBLAS on RISC-V CPUs, and 124% of cuBLAS on NVIDIA GPUs. Additionally, QiMeng-TensorOp also significantly reduces development costs by 200× compared with human experts.

The Chain-of-Thought (CoT) prompting mechanism has effectively enhanced the performance of large language models (LLMs) across a variety of natural language processing (NLP) tasks, including complex zero-shot learning scenarios. Recent studies suggest that this effectiveness arises from CoT's capacity to direct LLMs' attention toward task-relevant keywords. However, traditional CoT methods yield only marginal improvements in the realm of code generation, particularly for models with fewer than 10 billion parameters. We posit that this limitation stems from the substantial disparity between the logical structure and representational form of code compared to natural language. Considering the training and deployment costs, enhancing the performance of small LLMs through advanced prompting and instruction-tuning is essential. In this paper, we introduce COFT (Chain of Functional Triggers), a novel prompting strategy specifically designed for code generation tasks. The design of COFT is based on the following important observation: An optimal CoT tailored for code generation should clearly indicate the core functionality of each critical step, while employing standard identifiers prevalent within the coding domain. Extensive experiments conducted on representative small LLMs ($<10 \mathrm{B}$) benchmarks demonstrate that our COFT substantially outperforms vanilla CoT methods. In challenging zeroshot scenarios and the Pass@1 metric, COFT can improve the performance of fundation LLMs by up to 35.3 %. These empirical findings support our hypothesis that an appropriate design for CoT alongside instruction tuning can fully activate even smallersized LLMs, making them better zero-shot learners for code generation. The source code of COFT and the constructed instruction-tuning dataset will be released.

Random chunks generation attack is a more sophisticated form of side-channel attack, where a deduplication request containing one or more target chunks and several randomly generated non-duplicate chunks is able to be constructed. Once the number of required chunks or linear combinations in response is equal to that of the non-duplicate chunks, the existence privacy of target chunks is immediately exposed. Once such an attack is launched in a hybrid mode, which means some of the target chunks are replaced with public ones with known existence, the probability of privacy exposure further increases. In order to address these challenges, we propose a novel splicing strategy based chunk level deduplication scheme, which effectively protects the existence privacy of target chunks. Specifically, with the help of short hash functions, we introduce a secure splicing strategy for tags that divides query tags into multiple ternary tag groups in an unpredictable manner from the perspective of the attacker, which is the basis for chunk splicing. Furthermore, a response table is constructed accordingly to require uploading of the corresponding number of linear combinations based on the number of non-duplicate tag groups, which is able to introduce obfuscation in a lightweight way. Security analysis and experimental results demonstrate that the proposed scheme can effectively resist the aforementioned attacks with controllable overhead, regardless of the number of target chunks in request.

Abstract A collection of techniques is developed in this paper for ray casting free‐form deformed‐volume objects with high quality and efficiency. The known inverse ray deformation approach is combined with free‐form deformation to bend the rays to the opposite direction of the deformation, producing an image of the deformed volume without generating a really deformed intermediate volume. The local curvature is estimated and used for the adaptive selection of the length of polyline segments, which approximate the inversely deformed ray trajectories; thus longer polyline segments can be automatically selected in regions with small curvature, reducing deformation calculation without losing the spatial continuity of the simulated deformation. We developed an efficient method for the estimation of the local deformation function. The Jacobian of the local deformation function is used for adjustments of the opacity values and normal vectors computed from the original volume, guaranteeing that the deformed spatial structures are correctly rendered. The popular ray casting acceleration techniques, like early ray termination and space leaping, are incorporated into the deformation procedure, providing a speed‐up factor of 2.34–6.56 compared to the non‐optimized case. Copyright © 2003 John Wiley & Sons, Ltd.