Recommender systems have become ubiquitous in modern online services, yet their deployment across diverse hardware accelerators remains challenging due to significant performance variations. Contemporary deep learning recommendation models (DLRMs), such as DeepFM and NGCF, exhibit substantial inference latency differences when executed on NVIDIA GPUs,AMDGPUs, and Google TPUs, primarily due to architectural disparities and vendor-specific optimization strategies. Existing graph optimization frameworks are typically designed for specific hardware backends, lacking the flexibility to generate portable high-performance implementations across heterogeneous accelerators. This paper presents CAGR (Cross-Accelerator Graph Rewriting), a novel framework that achieves performance-portable inference optimization for recommendation models through three key innovations: (1) a hardware-aware graph rewriting engine that dynamically selects optimal operator implementations by analyzing compute-to-memory bandwidth ratios and operator density characteristics; (2) a reinforcement learning-based transformation policy that learns cross-platform optimization strategies without exhaustive search; and (3) a heterogeneous pipeline architecture enabling “optimize once, deploy across supported backends” semantics. We implement CAGR with support for multiple kernel backends including Triton, cuBLAS, MIOpen, and oneDNN, and demonstrate its effectiveness on the Avazu CTR prediction dataset. Experimental results show that CAGR achieves 1.8-3.2× speedup over baseline implementations across NVIDIA V100, AMD MI100, and Google TPU v3 platforms, while reducing optimization time by 67% compared to platform-specific auto-tuning approaches. Furthermore, CAGR maintains 92-96% of reference optimized performance with zero manual intervention, demonstrating practical viability for production recommendation systems requiring multi-vendor deployment.
Contemporary retrieval-augmented generation systems face a fundamental trade-off between semantic comprehensiveness and computational tractability when scaling to billion-token corpora. We present a unified framework that reconciles this tension through neural-symbolic dual-indexing, wherein sparse graph skeletons constructed from high-centrality document chunks enable structured reasoning, while complementary bipartite keyword indices ensure broad semantic coverage. Our architecture achieves this decomposition by formulating retrieval as a constrained optimization problem over heterogeneous index structures, employing Prize-Collecting Steiner Trees for subgraph extraction and Personalized PageRank for multi-hop traversal. Through synergistic integration of Graph Neural Networks with vector embeddings, the system performs explicit relational reasoning while maintaining sub-second query latency. Empirical evaluation across 6.0 benchmark datasets demonstrates that selective skeleton construction from the top 20.0% of chunks—identified via eigenvector centrality on k-nearest neighbor graphs—yields 10.0× cost reduction relative to exhaustive knowledge graph construction while improving generation quality by 32.4% and retrieval coverage by 92.4%. Furthermore, neural-symbolic coupling enables 7.0-billion parameter models to match GPT-4 performance on multi-hop question answering through single-step graph-guided inference, eliminating iterative retrieval overhead. Production deployments validate sub-200.0ms latency at scale through hierarchical caching strategies that reduce time-to-first-token by 4.0×. The proposed framework establishes dual-indexing as the canonical architecture for enterprise retrieval systems, providing a principled methodology for balancing semantic understanding against structured reasoning in large-scale information access.
Cloud-based intrusion detection systems (IDSs) increasingly rely on deep learning classifiers to identify malicious traffic; however, this reliance exposes them to adversarial evasion attacks in which adversaries craft near-imperceptible perturbations to bypass detection. Existing defenses based on conventional adversarial training often recover robustness against known perturbation patterns at the cost of degraded detection accuracy on canonical attack categories—a robustness–accuracy trade-off that remains an open challenge in the field. In this paper, we propose GT-CSAT (Game-Theoretic Cost-Sensitive Adversarial Training), a novel defense framework tailored for cloud security environments. GT-CSAT couples an improved Wasserstein GAN with Gradient Penalty (WGAN-GP) threat generator—conditioned on attack semantics to simulate functionally consistent and highly covert traffic variants—with a minimax adversarial training loop governed by a game-theoretic cost-sensitive loss function. The proposed loss function assigns asymmetric misclassification penalties derived from a two-player zero-sum payoff matrix, enabling the detector to maintain vigilance over both novel adversarial variants and well-characterized conventional threats simultaneously. Specifically, misclassifying an adversarially perturbed attack as benign incurs a strictly higher penalty than the symmetric cross-entropy baseline, while the cost weights are dynamically adapted via a Nash equilibrium-inspired update rule during training. We conduct comprehensive experiments on the Cloud Vulnerabilities Dataset (CVD), CICIDS-2017, and UNSW-NB15, which encompass diverse cloud-specific attack scenarios including denial-of-service, port scanning, brute-force, and SQL injection traffic. Under six representative evasion strategies—FGSM, PGD, C&W, BIM, DeepFool, and IDSGAN-style black-box perturbations—GT-CSAT achieves an average robust accuracy of 94.3%, surpassing standard adversarial training by 6.8 percentage points and the undefended baseline by 21.4 percentage points, while preserving clean-traffic detection at 97.1%. These results confirm that the game-theoretic cost structure effectively decouples robustness from accuracy, yielding a Pareto-superior detection profile relative to competing baselines across all evaluated threat models. The source code and experimental configurations have been publicly released to facilitate reproducibility.