TL;DR: A compact 2D convolutional neural network framework based on wavelet synchrosqueezing transform (WSST) time–frequency (TF) representations for EEG-based dementia classification demonstrates that the WSST-based representation provides an effective and robust input for CNN-based dementia classification under strict subject-disjoint validation.
Electroencephalography (EEG)-based discrimination of Alzheimer’s disease (AD), frontotemporal dementia (FTD), and cognitively normal controls (CN) remains challenging under clinically realistic subject-disjoint evaluation. This study presents a compact 2D convolutional neural network (CNN) framework based on wavelet synchrosqueezing transform (WSST) time–frequency (TF) representations for EEG-based dementia classification. Using a publicly available OpenNeuro dataset, the proposed framework was evaluated under strict subject-disjoint 10-fold cross-validation and compared against discrete wavelet transform (DWT) representations, adapted EEG-specific CNN baselines, and general ImageNet-pretrained transfer learning models. Across AD/CN, FTD/CN, and AD+FTD/CN tasks, the results show that WSST consistently outperforms DWT, achieving accuracies of 97.75%, 98.68%, and 98.10%, respectively, with substantially lower variability across folds. The proposed compact CNN also surpassed all EEG-specific and transfer learning baselines, indicating that architectures tailored to WSST-derived TF structure provide more stable and discriminative feature learning than models originally designed for raw EEG decoding or natural-image classification. Gradient-weighted Class Activation Mapping (Grad-CAM) analysis further provided a qualitative attribution check, indicating that higher-attribution regions are primarily located within structured WSST TF regions. These findings demonstrate that the WSST-based representation provides an effective and robust input for CNN-based dementia classification under strict subject-disjoint validation.
TL;DR: A sleep-inspired, replay-driven memory consolidation-based temporal learning framework that reconstructs context from replayed memory during brief sleep periods for temporal context learning in resource-constrained edge systems is proposed.
In edge computing applications, storing long temporal sequences is memory and energy intensive, while sensory data arrives sequentially, requiring online learning. Although online updates reduce storage requirements, they lack access to broader temporal context needed for accurat…
TL;DR: This work adopts a biologically plausible saliency-informed dropout technique as an explainable alternative to the unstructured standard random dropout approach and presents empirical results on regimes where such structured dynamic and static sparsities interact optimally to prune a ResNet model on the Imagenette dataset.
Biological vision has evolved to make efficient use of the limited information processing capability and tight energy budget of the brain by preferentially processing the most salient features of visual scenes. In contrast, modern deep vision models rely on expansive, high-dimens…
TL;DR: Shem is presented, a framework that enables gradient-based optimization of system-level objectives such as robustness and accuracy on user-defined analog compute models and introduces an autoparallelization algorithm that reduces total optimization runtime by 49–91%.
Analog compute paradigms are gaining attention for their potential to overcome the energy and latency limits of digital systems. Analog computations are modeled as dynamical systems for design and optimization, but this process is challenging: the dynamics are governed by nonline…
TL;DR: This work proposes heterogeneous neural networks that combine spiking neural networks (SNNs) and artificial neural networks (ANNs) at bandwidth-limited regions, such as chip boundaries, where spike-based communication reduces data transfer overhead.
Efficient communication is central to both biological and artificial intelligence (AI) systems. In biological brains, the challenge of long-range communication across regions is addressed through sparse, spike-based signaling, minimizing energy and latency. Conversely, modern AI…
TL;DR: The research rigorously investigates how neuromorphic architectures encode and integrate temporal information by conducting a comprehensive ablation study using a hybrid network, and demonstrates that shallow neuromorphic integration effectively maximizes the gains from temporal integration while mitigating the information loss inherent in binary spike quantization.
While neuromorphic systems offer a promising path for processing dynamic, event-based data, current benchmarks often fail to isolate the specific impact of temporal integration on model performance. To address this, our research rigorously investigates how neuromorphic architectu…
TL;DR: CRISP (Compiled RISP) is a neuromorphic simulator based on RISP, which compiles a given network into a minimal sequence of machine code for the desired platform entirely at runtime.
Simulators for spiking neural networks lie on the critical path between a researcher with an idea and a working network. Many of the current simulators have various implementation inefficiencies that hamper their performance across a wide variety of tasks. CRISP (Compiled RISP) i…