FuseFSS: Efficient Secure LLM Inference with Function Secret Sharing

• URL: http://arxiv.org/abs/2606.09551v1
• Date: Mon, 08 Jun 2026 14:30:00 GMT
• Title: FuseFSS: Efficient Secure LLM Inference with Function Secret Sharing
• Abstract summary: We present FuseFSS, a compiler that replaces per-operator protocol design with a single compilation pipeline.<n>Compared to the current state-of-the-art FSS-based GPU secure inference, FuseFSS preserves accuracy while achieving a $1.24times$--$1.50times$ end-to-end speedup.
• Score: 10.153726524753628
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Two-server secure inference allows a client to query a hosted large language model (LLM) without revealing prompts or embeddings. Recent GPU systems based on function secret sharing (FSS) make linear layers efficient, but fixed-point nonlinearities and helper operations remain a bottleneck because each operator is typically implemented as a bespoke protocol with its own comparisons, wrap-around corrections, and preprocessing material. We present FuseFSS, a compiler that replaces per-operator protocol design with a single compilation pipeline. For each scalar fixed-point operator, a compact specification lists its interval partition, low-degree arithmetic pieces, and required predicate bits. The compiler emits two batched FSS evaluations on the public masked value: one packed comparison that returns all predicate bits, and one vector interval lookup that returns the active coefficients and constants. Compared to the current state-of-the-art FSS-based GPU secure inference, FuseFSS preserves accuracy while achieving a $1.24\times$--$1.50\times$ end-to-end speedup and reducing online communication by $9\%$--$16\%$ on BERT and GPT-style models; preprocessing is also lighter, with $14\%$--$23\%$ lower key-generation time and $20\%$--$24\%$ smaller keys.

Related papers

• Towards Scalable Fuzzy PSI via Efficient Fuzzy Matching [43.14972106243527]
  We present fuzzy PSI protocols for general $L_p in [1, infty]$ distance, supporting both low- and high-dimensional sets.<n>The first is built from a role-reversed oblivious PRF (OPRF) and realizes $O(dlog )$ overhead, compared to $O(log )d)$ in previous works.<n>The second leverages customized oblivious transfer (OT) with $O(dell)$ overhead, where $ell$ is the bit length of inputs.
  arXiv  Detail & Related papers  (2026-08-12T00:30:53Z)
• SSTQ:Privacy-Preserving Vector Quantization via Subsampled Stochastic TurboQuant [79.24089819400126]
  Subsampled TurboQuant (SSTQ) is a framework that combines overcomplete equal-width tight frames, coordinate subsampling, and privacy-aware quantization.<n>SSTQ achieves mean squared error scaling while using only $lceil log N il + b$ bits per client, where $N = (d)$ is the frame size.<n>We also derive a surrogate privacy-aware codebook objective that reduces the codebook-dependent MSE scaling from $O(4b)$ to $O(2b)$.
  arXiv  Detail & Related papers  (2026-08-05T17:51:25Z)
• From Patches to Trajectories: Privileged Process Supervision for Software-Engineering Agents [56.31499185764872]
  Supervised fine-tuning (SFT) on long teacher trajectories is the dominant way to instill investigation and reasoning in open software-engineering (SWE) agents.<n>We propose Patches-to-Trajectories (P2T), which uses $pstar$ as privileged information during curation and formulates trajectory construction as bi-objective optimization over per-step effectiveness and trajectory length.
  arXiv  Detail & Related papers  (2026-05-21T04:54:55Z)
• Different Prompts, Different Ranks: Prompt-aware Dynamic Rank Selection for SVD-based LLM Compression [3.291633155351422]
  PARSE is a post-training framework for $textbfP$rompt-$textbfA$ware $textbfR$ank $textbfS$election as $textbfE$xperts in SVD-compressed LLMs.<n>It improves average task accuracy by up to 10% at a compression ratio of 0.6 on LLaMA-7B, and achieves up to 2.5 $times$ prefill and 2.4 $times$ decode speedup over native SVD execution.
  arXiv  Detail & Related papers  (2026-05-09T00:02:33Z)
• Sequential vs. Simultaneous Entanglement Swapping under Optimal Link-Layer Control [1.8829359083448898]
  Connection-less, packet-switched quantum network architectures distribute entanglement across multi-hop paths through sequential entanglement swapping.<n>We present a proof-of-principle study at fixed chain length $n = 4$ in which each link is governed by a fixed reinforcement-learning policy.<n>Sequence swapping collapses to zero end-to-end deliveries below $T_cmathrmext/= 25$, and begins recovering at $T_cmathrmext/= 50$.
  arXiv  Detail & Related papers  (2026-05-05T17:59:22Z)
• Efficient Arithmetic-and-Comparison Homomorphic Encryption with Space Switching [15.580543805873466]
  homomorphic encryption (FHE) enables computation on encrypted data without decryption.<n>No existing scheme efficiently supports both arithmetic and comparison operations in a unified framework.<n>We propose space switching method to integrate arithmetic and comparison seamlessly within FV-style schemes.
  arXiv  Detail & Related papers  (2026-04-21T18:12:37Z)
• Scaling Attention via Feature Sparsity [50.64995497733461]
  Scaling Transformers to ultra-long contexts is bottlenecked by the $O(n2 d)$ cost of self-attention.<n>We propose Sparse Feature Attention, where queries and keys are represented as $k$-sparse codes that preserve high-dimensional expressivity.<n>Across GPT-2 and Qwen3 pretraining, SFA matches dense baselines while improving speed by up to $2.5times$ and reducing FLOPs and KV-cache by nearly 50%.
  arXiv  Detail & Related papers  (2026-03-17T08:41:50Z)
• Privacy-Preserving Inference for Quantized BERT Models [13.36359444231145]
  Quantization offers a promising solution by converting floating-point operations into lower-precision integer computations.<n>We propose a fine-grained, layer-wise quantization scheme and support 1-bit weight fully connected layers in a secure setting.
  arXiv  Detail & Related papers  (2025-08-03T07:52:08Z)
• Fair Data Exchange at Near-Plaintext Efficiency [1.187519459637148]
  We introduce an FDE implementation that achieves near-plaintext speeds and sizes, making fair exchange practical even for gigabyte-scale files.<n>This can reduce transaction fees from roughly $10 to under $0.01 and shorten transaction latency from tens of seconds on down to about a second or less.
  arXiv  Detail & Related papers  (2025-06-17T19:50:25Z)
• FedSVD: Adaptive Orthogonalization for Private Federated Learning with LoRA [68.44043212834204]
  Low-Rank Adaptation (LoRA) is widely used for efficient fine-tuning of language models in learning (FL)<n>Low-Rank Adaptation (LoRA) is widely used for efficient fine-tuning of language models in learning (FL)
  arXiv  Detail & Related papers  (2025-05-19T07:32:56Z)
• MORSE: An Efficient Homomorphic Secret Sharing Scheme Enabling Non-Linear Operation [37.45826289062067]
  Homomorphic secret sharing (HSS) enables two servers to locally perform functions on encrypted data directly and obtain the results in the form of shares. A Paillier-based HSS solution seamlessly achieves multiplicative homomorphism and consumes less communication costs. We propose MORSE, an efficient homomorphic secret sharing scheme enabling non-linear operation.
  arXiv  Detail & Related papers  (2024-10-09T03:29:50Z)
• Rapid Person Re-Identification via Sub-space Consistency Regularization [51.76876061721556]
  Person Re-Identification (ReID) matches pedestrians across disjoint cameras. Existing ReID methods adopting real-value feature descriptors have achieved high accuracy, but they are low in efficiency due to the slow Euclidean distance computation. We propose a novel Sub-space Consistency Regularization (SCR) algorithm that can speed up the ReID procedure by 0.25$ times.
  arXiv  Detail & Related papers  (2022-07-13T02:44:05Z)
• Beta-CROWN: Efficient Bound Propagation with Per-neuron Split Constraints for Complete and Incomplete Neural Network Verification [151.62491805851107]
  We develop $beta$-CROWN, a bound propagation based verifier that can fully encode per-neuron splits. $beta$-CROWN is close to three orders of magnitude faster than LP-based BaB methods for robustness verification. By terminating BaB early, our method can also be used for incomplete verification.
  arXiv  Detail & Related papers  (2021-03-11T11:56:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.