Papers for

analog hardware engineers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Analog oscillatory hardware learns with lock-in equilibrium propagation

Lock-in EP: An In-Situ Training Algorithm for Oscillatory Hardware

Abstract: Analog hardware platforms offer the potential to reduce energy consumption over digital architectures, but in order to succeed, large-scale analog systems must also be able to operate with or recover from the variability of their components. Towards this goal, we derive and demonstrate the lock-in equilibrium propagation (LIEP) training method. LIEP provides local gradient information for each component in an oscillatory network without separate forward and backward sweeps, potentially allowing for in-situ learning capabilities on analog oscillatory hardware platforms. We demonstrate that LIEP can be used both for ab-initio training as well as recovering performance when pre-trained parameters are perturbed. We show that LIEP can be formulated as a three-factor update rule, and suggest that although the method is currently only validated on shallow networks, alternate architectures may allow it to extend to deep and large-scale networks addressing complex tasks.

Mon 5 OctMachine Learning

The gist

Analog computers could use less energy than digital ones, but their parts aren’t always reliable. The authors developed a training method called lock-in equilibrium propagation (LIEP) that helps analog systems learn by updating themselves without extra steps. This method can train networks from scratch or fix them after their settings get changed. Their approach may help analog devices learn directly on the hardware, although it has so far only been tested on simple networks.

Open → 2610.07283v1

Predictive coding method works without activation function derivatives

Analog-Friendly Predictive Coding without Activation Derivatives

Abstract: Predictive coding (PC) is a local, energy-based alternative to backpropagation (BP) whose iterative inference dynamics make it attractive for implementation on analog hardware. However, standard nonlinear PC requires evaluating the derivative of the activation function during both inference and learning, which can be difficult to realise physically. Here, we introduce \textit{activation-matched Bregman PC}, replacing standard squared-error energies with Bregman divergences matched to the activation function. This formulation eliminates activation derivatives and, when combined with inference via mirror descent, yields local inference and learning rules requiring only weighted sums, local prediction errors, state integration, and the activation function. In digital experiments, Bregman PC performs comparably to standard PC and BP on classification and generative tasks, while preserving characteristic learning dynamics of PC and its convergence to BP under stable large-model parameterisations. These results provide a more analog-friendly formulation of nonlinear PC while retaining its key computational properties.

Sat 26 SeptMachine LearningArtificial IntelligenceNeural and Evolutionary Computing

The gist

Training neural networks usually requires calculating how outputs change with inputs, which can be hard to do on analog devices. The authors introduce a new way to train predictive coding models that skips this tricky step by using a different mathematical approach tied to the activation function. This makes it easier to run these models on analog hardware without losing their ability to learn well. They tested their method on tasks like classification and generating data, finding it works about as well as traditional methods.

Open → 2609.32350v1