Atsushi Uchida

Ryugo Iwami, Takatomo Mihana, Kazutaka Kanno et al.

Photonic artificial intelligence has attracted considerable interest in accelerating machine learning; however, the unique optical properties have not been fully utilized for achieving higher-order functionalities. Chaotic itinerancy, with its spontaneous transient dynamics among multiple quasi-attractors, can be employed to realize brain-like functionalities. In this paper, we propose a method for controlling the chaotic itinerancy in a multi-mode semiconductor laser to solve a machine learning task, known as the multi-armed bandit problem, which is fundamental to reinforcement learning. The proposed method utilizes ultrafast chaotic itinerant motion in mode competition dynamics controlled via optical injection. We found that the exploration mechanism is completely different from a conventional searching algorithm and is highly scalable, outperforming the conventional approaches for large-scale bandit problems. This study paves the way to utilize chaotic itinerancy for effectively solving complex machine learning tasks as photonic hardware accelerators.

Kensei Morijiri, Kento Takehana, Takatomo Mihana et al.

Photonic accelerators have attracted increasing attention in artificial intelligence applications. The multi-armed bandit problem is a fundamental problem of decision making using reinforcement learning. However, the scalability of photonic decision making has not yet been demonstrated in experiments, owing to technical difficulties in physical realization. We propose a parallel photonic decision-making system for solving large-scale multi-armed bandit problems using optical spatiotemporal chaos. We solve a 512-armed bandit problem online, which is much larger than previous experiments by two orders of magnitude. The scaling property for correct decision making is examined as a function of the number of slot machines, evaluated as an exponent of 0.86. This exponent is smaller than that in previous work, indicating the superiority of the proposed parallel principle. This experimental demonstration facilitates photonic decision making to solve large-scale multi-armed bandit problems for future photonic accelerators.

Takashi Urushibara, Nicolas Chauvet, Satoshi Kochi et al.

Accelerating artificial intelligence by photonics is an active field of study aiming to exploit the unique properties of photons. Reinforcement learning is an important branch of machine learning, and photonic decision-making principles have been demonstrated with respect to the multi-armed bandit problems. However, reinforcement learning could involve a massive number of states, unlike previously demonstrated bandit problems where the number of states is only one. Q-learning is a well-known approach in reinforcement learning that can deal with many states. The architecture of Q-learning, however, does not fit well photonic implementations due to its separation of update rule and the action selection. In this study, we organize a new architecture for multi-state reinforcement learning as a parallel array of bandit problems in order to benefit from photonic decision-makers, which we call parallel bandit architecture for reinforcement learning or PBRL in short. Taking a cart-pole balancing problem as an instance, we demonstrate that PBRL adapts to the environment in fewer time steps than Q-learning. Furthermore, PBRL yields faster adaptation when operated with a chaotic laser time series than the case with uniformly distributed pseudorandom numbers where the autocorrelation inherent in the laser chaos provides a positive effect. We also find that the variety of states that the system undergoes during the learning phase exhibits completely different properties between PBRL and Q-learning. The insights obtained through the present study are also beneficial for existing computing platforms, not just photonic realizations, in accelerating performances by the PBRL algorithms and correlated random sequences.

Felix Köster, Atsushi Uchida

Physical dynamical systems can be viewed as natural information processors: their systems preserve, transform, and disperse input information. This perspective motivates learning not only from data generated by such systems, but also how to measure them in a way that extracts the most useful information for a given task. We propose a general computing framework for adaptive information extraction from dynamical systems, in which a trainable attention module learns both where to probe the system state and how to combine these measurements to optimize prediction performance. As a concrete instantiation, we implement this idea using a spatiotemporal field governed by a partial differential equation as the underlying dynamics, though the framework applies equally to any system whose state can be sampled. Our results show that adaptive spatial sensing significantly improves prediction accuracy on canonical chaotic benchmarks. This work provides a perspective on attention-enhanced reservoir computing as a special case of a broader paradigm: neural networks as trainable measurement devices for extracting information from physical dynamical systems.

Felix Köster, Kazutaka Kanno, Jun Ohkubo et al.

Photonic reservoir computing has been successfully utilized in time-series prediction as the need for hardware implementations has increased. Prediction of chaotic time series remains a significant challenge, an area where the conventional reservoir computing framework encounters limitations of prediction accuracy. We introduce an attention mechanism to the reservoir computing model in the output stage. This attention layer is designed to prioritize distinct features and temporal sequences, thereby substantially enhancing the prediction accuracy. Our results show that a photonic reservoir computer enhanced with the attention mechanism exhibits improved prediction capabilities for smaller reservoirs. These advancements highlight the transformative possibilities of reservoir computing for practical applications where accurate prediction of chaotic time series is crucial.

Satoshi Sunada, Tomoaki Niiyama, Kazutaka Kanno et al.

The rapidly increasing computational demands for artificial intelligence (AI) have spurred the exploration of computing principles beyond conventional digital computers. Physical neural networks (PNNs) offer efficient neuromorphic information processing by harnessing the innate computational power of physical processes; however, training their weight parameters is computationally expensive. We propose a training approach for substantially reducing this training cost. Our training approach merges an optimal control method for continuous-time dynamical systems with a biologically plausible training method--direct feedback alignment. In addition to the reduction of training time, this approach achieves robust processing even under measurement errors and noise without requiring detailed system information. The effectiveness was numerically and experimentally verified in an optoelectronic delay system. Our approach significantly extends the range of physical systems practically usable as PNNs.

Tomoki Yamagami, Etsuo Segawa, Takatomo Mihana et al.

Quantum reinforcement learning has emerged as a framework combining quantum computation with sequential decision-making, and applications to the multi-armed bandit (MAB) problem have been reported. The graph bandit problem extends the MAB setting by introducing spatial constraints, yet quantum approaches remain limited. We propose a quantum algorithm for best-arm identification in graph bandits, termed Quantum Spatial Best-Arm Identification (QSBAI). The method employs quantum walks to encode superpositions over graph-constrained actions, extending amplitude amplification and generalizing the Quantum BAI algorithm via Szegedy's walk framework. This establishes a link between Grover-type search and reinforcement learning tasks with structural restrictions. We analyze complete and bipartite graphs, deriving the maximal success probability of identifying the best arm and the time step at which it is achieved. Our results highlight the potential of quantum walks to accelerate exploration in constrained environments and extend the applicability of quantum algorithms for decision-making.

Felix Köster, Atsushi Uchida

Large Language Models (LLM) have dominated the science and media landscape duo to their impressive performance on processing large chunks of data and produce human-like levels of text. Nevertheless, their huge energy demand and slow processing still a bottleneck for further increasing quality while also making the models accessible to everyone. To solve this bottleneck, we will investigate how reservoir computing performs on natural text processing, which could enable fast and energy efficient hardware implementations. Studies investigating the use of reservoir computing as a language model remain sparse. In this paper, we compare three distinct approaches for character-level language modeling, two different reservoir computing approaches, where only an output layer is trainable, and the well-known transformer-based architectures, which fully learn an attention-based sequence representation. We explore the performance, computational cost and prediction accuracy for both paradigms by equally varying the number of trainable parameters for all models. Using a consistent pipeline for all three approaches, we demonstrate that transformers excel in prediction quality, whereas reservoir computers remain highly efficient reducing the training and inference speed. Furthermore, we investigate two types of reservoir computing: a traditional reservoir with a static linear readout, and an attention-enhanced reservoir that dynamically adapts its output weights via an attention mechanism. Our findings underline how these paradigms scale and offer guidelines to balance resource constraints with performance.

Satoshi Sunada, Atsushi Uchida

Photonic neural networks have significant potential for high-speed neural processing with low latency and ultralow energy consumption. However, the on-chip implementation of a large-scale neural network is still challenging owing to its low scalability. Herein, we propose the concept of a photonic neural field and implement it experimentally on a silicon chip to realize highly scalable neuro-inspired computing. In contrast to existing photonic neural networks, the photonic neural field is a spatially continuous field that nonlinearly responds to optical inputs, and its high spatial degrees of freedom allow for large-scale and high-density neural processing on a millimeter-scale chip. In this study, we use the on-chip photonic neural field as a reservoir of information and demonstrate a high-speed chaotic time-series prediction with low errors using a training approach similar to reservoir computing. We discuss that the photonic neural field is potentially capable of executing more than one peta multiply-accumulate operations per second for a single input wavelength on a footprint as small as a few square millimeters. In addition to processing, the photonic neural field can be used for rapidly sensing the temporal variation of an optical phase, facilitated by its high sensitivity to optical inputs. The merging of optical processing with optical sensing paves the way for an end-to-end data-driven optical sensing scheme.

Kazutaka Kanno, Makoto Naruse, Atsushi Uchida

Photonic reservoir computing is an emergent technology toward beyond-Neumann computing. Although photonic reservoir computing provides superior performance in environments whose characteristics are coincident with the training datasets for the reservoir, the performance is significantly degraded if these characteristics deviate from the original knowledge used in the training phase. Here, we propose a scheme of adaptive model selection in photonic reservoir computing using reinforcement learning. In this scheme, a temporal waveform is generated by different dynamic source models that change over time. The system autonomously identifies the best source model for the task of time series prediction using photonic reservoir computing and reinforcement learning. We prepare two types of output weights for the source models, and the system adaptively selected the correct model using reinforcement learning, where the prediction errors are associated with rewards. We succeed in adaptive model selection when the source signal is temporally mixed, having originally been generated by two different dynamic system models, as well as when the signal is a mixture from the same model but with different parameter values. This study paves the way for autonomous behavior in photonic artificial intelligence and could lead to new applications in load forecasting and multi-objective control, where frequent environment changes are expected.

Tomoaki Niiyama, Genki Furuhata, Atsushi Uchida et al.

Decision making is a fundamental capability of living organisms, and has recently been gaining increasing importance in many engineering applications. Here, we consider a simple decision-making principle to identify an optimal choice in multi-armed bandit (MAB) problems, which is fundamental in the context of reinforcement learning. We demonstrate that the identification mechanism of the method is well described by using a competitive ecosystem model, i.e., the competitive Lotka--Volterra (LV) model. Based on the "winner-take-all" mechanism in the competitive LV model, we demonstrate that non-best choices are eliminated and only the best choice survives; the failure of the non-best choices exponentially decreases while repeating the choice trials. Furthermore, we apply a mean-field approximation to the proposed decision-making method and show that the method has an excellent scalability of $O(\log N)$ with respect to the number of choices $N$. These results allow for a new perspective on optimal search capabilities in competitive systems.

Makoto Naruse, Takashi Matsubara, Nicolas Chauvet et al.

Generative adversarial network (GAN) is gaining increased importance in artificially constructing natural images and related functionalities wherein two networks called generator and discriminator are evolving through adversarial mechanisms. Using deep convolutional neural networks and related techniques, high-resolution, highly realistic scenes, human faces, among others have been generated. While GAN in general needs a large amount of genuine training data sets, it is noteworthy that vast amounts of pseudorandom numbers are required. Here we utilize chaotic time series generated experimentally by semiconductor lasers for the latent variables of GAN whereby the inherent nature of chaos can be reflected or transformed into the generated output data. We show that the similarity in proximity, which is a degree of robustness of the generated images with respects to a minute change in the input latent variables, is enhanced while the versatility as a whole is not severely degraded. Furthermore, we demonstrate that the surrogate chaos time series eliminates the signature of generated images that is originally observed corresponding to the negative autocorrelation inherent in the chaos sequence. We also discuss the impact of utilizing chaotic time series in retrieving images from the trained generator.

Makoto Naruse, Takatomo Mihana, Hirokazu Hori et al.

Reinforcement learning involves decision making in dynamic and uncertain environments and constitutes a crucial element of artificial intelligence. In our previous work, we experimentally demonstrated that the ultrafast chaotic oscillatory dynamics of lasers can be used to solve the two-armed bandit problem efficiently, which requires decision making concerning a class of difficult trade-offs called the exploration-exploitation dilemma. However, only two selections were employed in that research; thus, the scalability of the laser-chaos-based reinforcement learning should be clarified. In this study, we demonstrated a scalable, pipelined principle of resolving the multi-armed bandit problem by introducing time-division multiplexing of chaotically oscillated ultrafast time-series. The experimental demonstrations in which bandit problems with up to 64 arms were successfully solved are presented in this report. Detailed analyses are also provided that include performance comparisons among laser chaos signals generated in different physical conditions, which coincide with the diffusivity inherent in the time series. This study paves the way for ultrafast reinforcement learning by taking advantage of the ultrahigh bandwidths of light wave and practical enabling technologies.

Makoto Naruse, Yuta Terashima, Atsushi Uchida et al.

Reinforcement learning involves decision making in dynamic and uncertain environments, and constitutes one important element of artificial intelligence (AI). In this paper, we experimentally demonstrate that the ultrafast chaotic oscillatory dynamics of lasers efficiently solve the multi-armed bandit problem (MAB), which requires decision making concerning a class of difficult trade-offs called the exploration-exploitation dilemma. To solve the MAB, a certain degree of randomness is required for exploration purposes. However, pseudo-random numbers generated using conventional electronic circuitry encounter severe limitations in terms of their data rate and the quality of randomness due to their algorithmic foundations. We generate laser chaos signals using a semiconductor laser sampled at a maximum rate of 100 GSample/s, and combine it with a simple decision-making principle called tug-of-war with a variable threshold, to ensure ultrafast, adaptive and accurate decision making at a maximum adaptation speed of 1 GHz. We found that decision-making performance was maximized with an optimal sampling interval, and we highlight the exact coincidence between the negative autocorrelation inherent in laser chaos and decision-making performance. This study paves the way for a new realm of ultrafast photonics in the age of AI, where the ultrahigh bandwidth of photons can provide new value.