🤖 AI Summary
Existing spiking neural networks struggle to simultaneously achieve trainability, dynamic diversity, and activity sparsity in temporal regression tasks, often suffering from discretization errors and noise sensitivity. This work proposes a differentiable spiking neuron model based on multi-timescale conductances, which integrates fast, slow, and ultraslow conductance dynamics to jointly modulate current–voltage characteristics. Within a unified architecture, the model naturally supports diverse firing patterns—including tonic, phasic, and burst spiking—while enabling end-to-end gradient-based learning without surrogate gradients. The design is both circuit-realizable and exhibits controllable excitability. Evaluated on the Mackey–Glass time-series prediction task, the proposed model significantly outperforms standard LIF and AdLIF neurons, achieving higher prediction accuracy with substantially lower spike activity density.
📝 Abstract
Spiking neural networks (SNNs) promise low-power event-driven computation for temporally rich tasks, but commonly used neuron models often trade off gradient-based trainability, dynamical richness, and high activity sparsity. These limitations are acute in regression, where approximation error, noise and spike discretization can severely degrade continuous-valued outputs. Indeed, many state-of-the-art (SOTA) SNNs rely on simple phenomenological dynamics trained with surrogate gradients and offer limited control over spiking diversity and sparsity. To overcome such limitations, we introduce multi-timescale conductance spiking networks, a gradient-trainable framework in which neural dynamics emerge from shaping the current-voltage (I-V) curve by tuning fast, slow and ultra-slow conductances. This parametrization allows systematic control over excitability, can be implemented efficiently in analog circuits, and yields rich firing regimes including tonic, phasic and bursting responses within a single model. We derive a discrete-time formulation of these differentiable dynamics, enabling direct backpropagation through time without surrogate-gradient approximations. To probe both trainability and accuracy, we evaluate feedforward networks of these neurons at the predictability limit of Mackey-Glass time-series regression and compare them to baseline LIF and SOTA AdLIF networks. Our model outperforms LIF and AdLIF networks, while exhibiting substantially sparser activity from both communication and computational perspectives. These results highlight multi-timescale conductance spiking neurons as a promising building block for energy-aware temporal processing and neuromorphic implementation.