Modality-Dependent Memory Mechanisms in Cross-Modal Neuromorphic Computing
This work addresses the challenge of optimizing memory mechanisms for energy-efficient neuromorphic computing across different sensory modalities, though it is incremental as it builds on existing methods with a systematic evaluation.
The paper tackled the problem of memory mechanisms in spiking neural networks (SNNs) lacking cross-modal generalization, revealing modality-dependent performance patterns with Hopfield networks showing a 21.53% accuracy gap between visual and auditory tasks, while supervised contrastive learning reduced this gap to 14.56% and achieved up to 603x energy efficiency over traditional networks.
Memory-augmented spiking neural networks (SNNs) promise energy-efficient neuromorphic computing, yet their generalization across sensory modalities remains unexplored. We present the first comprehensive cross-modal ablation study of memory mechanisms in SNNs, evaluating Hopfield networks, Hierarchical Gated Recurrent Networks (HGRNs), and supervised contrastive learning (SCL) across visual (N-MNIST) and auditory (SHD) neuromorphic datasets. Our systematic evaluation of five architectures reveals striking modality-dependent performance patterns: Hopfield networks achieve 97.68% accuracy on visual tasks but only 76.15% on auditory tasks (21.53 point gap), revealing severe modality-specific specialization, while SCL demonstrates more balanced cross-modal performance (96.72% visual, 82.16% audio, 14.56 point gap). These findings establish that memory mechanisms exhibit task-specific benefits rather than universal applicability. Joint multi-modal training with HGRN achieves 94.41% visual and 79.37% audio accuracy (88.78% average), matching parallel HGRN performance through unified deployment. Quantitative engram analysis confirms weak cross-modal alignment (0.038 similarity), validating our parallel architecture design. Our work provides the first empirical evidence for modality-specific memory optimization in neuromorphic systems, achieving 603x energy efficiency over traditional neural networks.