In situ fine-tuning of in silico trained Optical Neural Networks

Gianluca Kosmella; Ripalta Stabile; Jaron Sanders

Home/Archive/arXiv

arXiv · 2506.22122

In situ fine-tuning of in silico trained Optical Neural Networks

Gianluca Kosmella·Ripalta Stabile·Jaron Sanders

View original source ↗Read source document ↗Export metadata ↓

Abstract

Optical Neural Networks (ONNs) promise significant advantages over traditional electronic neural networks, including ultrafast computation, high bandwidth, and low energy consumption, by leveraging the intrinsic capabilities of photonics. However, training ONNs poses unique challenges, notably the reliance on simplified in silico models whose trained parameters must subsequently be mapped to physical hardware. This process often introduces inaccuracies due to discrepancies between the idealized digital model and the physical ONN implementation, particularly stemming from noise and fabrication imperfections. In this paper, we analyze how noise misspecification during in silico training impacts ONN performance and we introduce Gradient-Informed Fine-Tuning (GIFT), a lightweight algorithm designed to mitigate this performance degradation. GIFT uses gradient information derived from the noise structure of the ONN to adapt pretrained parameters directly in situ, without requiring expensive retraining or complex experimental setups. GIFT comes with formal conditions under which it improves ONN performance. We also demonstrate the effectiveness of GIFT via simulation on a five-layer feed forward ONN trained on the MNIST digit classification task. GIFT achieves up to $28\%$ relative accuracy improvement compared to the baseline performance under noise misspecification, without resorting to costly retraining. Overall, GIFT provides a practical solution for bridging the gap between simplified digital models and real-world ONN implementations.

Explore related subjects

cs.NE cs.ET eess.SP

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Gianluca Kosmella, Ripalta Stabile, Jaron Sanders. 2025-06-27. In situ fine-tuning of in silico trained Optical Neural Networks. https://arxiv.org/abs/2506.22122

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Rethinking Pairwise Token Interaction in Spiking Transformers

Spiking Transformers inherit token interaction mechanisms from conventional Transformers, yet their sparse binary representations fundamentally alter how token-to-token communication is established. In particular, spike-based query-key matching produces highly sparse and input-dependent interaction patterns, coupling information propagation to the instantaneous availability of matching spike events. This motivates a different interaction paradigm in which long-range communication does not rely solely on pairwise spike coincidence. We therefore propose Gated Spike Axial Propagation (GSAP), a spike-native token interaction mechanism that decouples information propagation from context selection. Instead of directly determining communication through query-key matching, GSAP first propagates spike-based context along the horizontal and vertical axes, allowing information to reach distant tokens through structured sequential propagation. A receiver-conditioned gate then determines how much of the propagated context is incorporated at each token, while a lightweight local pathway preserves fine-grained neighborhood information. In this way, GSAP reformulates token interaction as a propagate-then-select process, enabling structured long-range communication while retaining the sparse event-driven nature of spiking representations. Code is available at https://github.com/Fancyssc/GSAP.

cs.NE↗

arXiv2026

NeuroRule: Making Black-Box Neural Networks Explainable through Rule-set Evolution

High-capacity neural network models have achieved state-of-the-art performance across diverse classification tasks, yet they frequently operate as black-box models, lacking the transparency necessary for critical decision-making. Such opacity creates a persistent trade-off between performance and explainability. This paper proposes a solution to address this gap: the NeuroRule knowledge distillation framework that results in explainable rule-sets from neural network models. NeuroRule adapts the EVOTER rule-set evolution infrastructure to treat neural networks as targets for the evolution process, distilling their performance into concise sets of propositional logic expressions. There are three primary contributions: (1) an evolutionary method for distilling black-box neural network models into explicit rule-set models; (2) a method for making rule sets more explainable by including a conciseness objective to evolution; and (3) a demonstration that the distillation is viable even without access to the original neural network training data. The paper thus establishes that black-box neural network models can be made explainable and therefore useful in real-world applications where trustworthiness is paramount.

cs.NE↗

arXiv2026

The Computational Value of Sensory-Aligned Receptive Fields Depends on Neuronal Expressivity

Biological sensory neurons have selective receptive fields organized along meaningful stimulus coordinates, such as frequency, motion direction, or retinotopic position. Such structure may arise from efficient coding and biological constraints on activity, connectivity, and wiring, as computational studies of simple neurons have shown across modalities. This raises a question: do structured receptive fields confer a computational advantage beyond resource efficiency itself, and does this advantage persist when individual neurons are highly expressive? We address this question in recurrent networks of Expressive Leaky Memory neurons, where we can independently vary neuronal complexity and the organization of feed-forward receptive fields. Across auditory and event-based visual classification tasks, receptive fields aligned with a task-relevant sensory coordinate improve test accuracy relative to budget-matched random receptive fields. This advantage disappears when sensory coordinates are scrambled, or when receptive fields follow task-irrelevant coordinates, showing that the benefit comes from alignment with task geometry rather than restricted connectivity alone. Increasing neuronal complexity reduces the performance advantage of structured receptive fields. Finally, generic synaptic sparsity regularization induces input selectivity and partially recovers performance, but remains substantially below explicitly structured receptive fields, suggesting that sparsity alone is insufficient to recover the full computational benefit of task-aligned receptive fields. Together, our results show that appropriate receptive fields can serve as a computational prior beyond sparsity itself, and that their value depends on the computational expressivity of individual neurons.

cs.NE↗