Search papers, labs, and topics across Lattice.
This paper introduces Analog Interaction Systems (AIS), a framework that reconciles the fixed dynamics of analog hardware with the flexibility required for effective generative modeling. By employing time-varying piecewise parameters and hidden physical states, the authors demonstrate a significant reduction in the expressivity gap compared to neural network baselines. The proposed approach achieves a remarkable energy efficiency of 23 碌J per generated image while outperforming existing analog generative models by 3-4 times on standard datasets like MNIST and Fashion-MNIST.
Achieving a 2-orders-of-magnitude energy efficiency improvement, the new AIS framework outperforms prior analog generative models by up to 4x.
Analog hardware platforms such as coupled oscillators and Analog Ising Machines naturally solve differential equations at a fraction of the energy cost of digital computation, making them attractive for low-power generative modeling, yet a fundamental mismatch exists: modern generative models assume flexible, software-defined dynamics, whereas analog hardware imposes fixed, physics-determined differential equations with limited approximation capacity. This paper introduces Analog Interaction Systems (AIS), a unified framework for hardware-implementable dynamical systems, and empirically characterizes their expressivity gap relative to neural network baselines. Two hardware-compatible mechanisms are proposed to narrow this gap - time-varying piecewise parameters and hidden physical states - and a Wasserstein GAN training procedure is developed to enable training of these models without requiring them to follow a specific trajectory. We characterize how area and power scale with connection density and precision, showing that sparse connectivity and low-bit-width quantized parameters are necessary for practical implementation, and estimate an energy cost of 23uJ per generated image for the chosen architecture, representing a 2-orders-of-magnitude improvement over digital baselines. On MNIST and Fashion-MNIST, our oscillator-based AIS achieves FID scores of 27.6 and 80.8, outperforming the best prior hardware-implementable analog generative models by 3-4x with a 4-bit sparse architecture.
Stanford HAI