Lune

HPCA2025Top-tier venue

CogSys: Efficient and Scalable Neurosymbolic Cognition System via Algorithm-Hardware Co-Design

Zishen Wan, Hanchen Yang, Ritik Raj, Che-Kai Liu, Ananda Samajdar, Arijit Raychowdhury, Tushar Krishna

Abstract

Neurosymbolic AI is an emerging compositional paradigm that fuses neural learning with symbolic reasoning to enhance the transparency, interpretability, and trustworthiness of AI. It also exhibits higher data efficiency making it promising for edge deployments. Despite the algorithmic promises and demonstrations, unfortunately executing neurosymbolic workloads on current hardware (CPU/GPU/TPU) is challenging due to higher memory intensity, greater compute heterogeneity and access pattern irregularity, leading to severe hardware underutilization. This work proposes CogSys, a characterization and co-design framework dedicated to neurosymbolic AI system acceleration, aiming to win both reasoning efficiency and scalability. On the algorithm side, CogSys proposes an efficient factorization technique to alleviate compute and memory overhead. On the hardware side, CogSys proposes a scalable neurosymbolic architecture with reconfigurable neuro/symbolic processing elements (nsPEn s P E) and bubble streaming (BS) dataflow with spatial-temporal (ST) mapping for highly parallel and efficient neurosymbolic computation. On the system side, CogSys features an adaptive workload-aware scheduler (adSCH) to orchestrate heterogeneous kernels and enhance resource utilization. Evaluated across cognitive workloads, CogSys enables reconfigurable support for neural and symbolic kernels and exhibits >75×\gt75 \times speedup over TPU-like systolic array with only <5%\lt5 \% area overhead, as benchmarked under the TSMC 28nm technology node. CogSys achieves 4×4 \times 96×96 \times speedup compared to desktop and edge GPUs. For the first time, CogSys enables real-time abduction reasoning towards human fluid intelligence, requiring only 0.3 s per reasoning task with 4 mm24 \mathrm{~mm}^{2} area and 1.48 W power consumption.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

2 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • REASON: Accelerating Probabilistic Logical Reasoning for Scalable Neuro-Symbolic Intelligence cites NSFlow: An End-to-End FPGA Framework with Scalable Dataflow Architecture for Neuro-Symbolic AI
  • NSFlow: An End-to-End FPGA Framework with Scalable Dataflow Architecture for Neuro-Symbolic AI cites CLEVRER: Collision Events for Video Representation and Reasoning
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers2

  1. NSFlow: An End-to-End FPGA Framework with Scalable Dataflow Architecture for Neuro-Symbolic AIHanchen Yang, Zishen Wan, Ritik Raj, Joongun Park et al.DAC 2025 · 8 citations
  2. REASON: Accelerating Probabilistic Logical Reasoning for Scalable Neuro-Symbolic IntelligenceZishen Wan, Che-Kai Liu, Jiayi Qian, Hanchen Yang et al.HPCA 2026 · 2 citations

Builds on25

  1. Learning Transferable Visual Models From Natural Language SupervisionAlec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh et al.ICML 2021 · 47,906 citations
  2. Long Range Arena : A Benchmark for Efficient TransformersYi Tay, Mostafa Dehghani, Samira Abnar, Yikang Shen et al.ICLR 2021 · 881 citations
  3. CLEVRER: Collision Events for Video Representation and ReasoningKexin Yi, Chuang Gan, Yunzhu Li, Pushmeet Kohli et al.ICLR 2020 · 584 citations
  4. Gemmini: Enabling Systematic Deep-Learning Architecture Evaluation via Full-Stack IntegrationHasan Genc, Seah Kim, Alon Amid, Ameer Haj-Ali et al.DAC 2021 · 325 citations
  5. F1: A Fast and Programmable Accelerator for Fully Homomorphic EncryptionNikola Samardzic, Axel Feldmann, Aleksandar Krastev, Srinivas Devadas et al.MICRO 2021 · 294 citations

Related papers

  1. Lobster: A GPU-Accelerated Framework for Neurosymbolic ProgrammingPaul Biberstein, Ziyang Li, Joseph Devietti, Mayur NaikASPLOS 2026 · 1 citation
  2. Compositional AI Beyond LLMs: System Implications of Neuro-Symbolic-Probabilistic ArchitecturesZishen Wan, Hanchen Yang, Jiayi Qian, Ritik Raj et al.ASPLOS 2026 · 2 citations
  3. FactorHD: A Hyperdimensional Computing Model for Multi-Object Multi-Class Representation and FactorizationYifei Zhou, Xuchu Huang, Chenyu Ni, Min Zhou et al.DAC 2025 · 1 citation
  4. KLay: Accelerating Arithmetic Circuits for Neurosymbolic AIJaron Maene, Vincent Derkinderen, Pedro Zuidberg Dos MartiresICLR 2025
  5. DOLPHIN: A Programmable Framework for Scalable Neurosymbolic LearningAaditya Naik, Jason Liu, Claire Wang, Amish Sethi et al.ICML 2025
Dusk over the sea between two cliffs drawn in fine vertical lines