![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
- Joined
- Dec 19, 2005
- Messages
- 18,300
"Reservoir computing based on solitary-like waves dynamics of liquid film flows: A proof of concept"
"Several theoretical works have shown that solitons---waves that self-maintain constant shape and velocity as they propagate---can be used as a physical computational reservoir, a concept where machine learning algorithms designed for digital computers are replaced by analog physical systems that exhibit nonlinear dynamical behaviour. Here we propose and experimentally validate a novel reservoir computing (RC) system that for the first time employs solitary-like (SL) waves propagating on the surface of a liquid film flowing over an inclined surface. We demonstrate the ability of the SL wave RC system (SLRC) to forecast chaotic time series, also producing experimental evidence of the possibility to combine RC with nonlinear vector autoregression techniques not only in a computer program but also in a physical systems."
---
""Reservoir computing (RC) has evolved significantly over the years as a branch of recurrent neural networks (RNNs) with a uniquely simplified training strategy. The core idea of RC is to take advantage of the inherent dynamics of a high-dimensional system, referred to as the "reservoir," to process temporal information. This paradigm shift has facilitated the exploration of various physical systems as reservoirs, and among them, solitary-like (SL) waves present in liquid film flows over inclined surfaces stand out.
SL waves, a type of soliton wave, maintain their shape and velocity as they travel. They naturally exist in many physical systems, including fluid flows. These waves can be thought of as individual information carriers, each propagating through the system according to its internal dynamics. The hypothesis is that the interaction and evolution of these waves, due to their nonlinear nature, can be harnessed to perform complex computations.
In our study, we designed and validated a SL wave-based RC system (SLRC). We accomplished this by creating a setup where a liquid film flowed over an inclined surface, producing SL waves. We then implemented an input method to perturb the system and a readout mechanism to capture the resulting wave dynamics, thus forming the SLRC.
The most significant highlight of this work is the successful application of SLRC for predicting chaotic time series. Chaotic systems, by definition, are highly sensitive to initial conditions, and their future states can be challenging to predict accurately. However, the nonlinear dynamic behavior exhibited by the SL waves provides a powerful tool to tackle this issue. The results showed that the SLRC could learn and forecast the complex patterns of these systems effectively.
Additionally, we demonstrated the capability of integrating the SLRC with nonlinear vector autoregression techniques. Traditionally, these techniques were used only in digital computation, but here we showed that they could be applied in the physical realm. This integration allows us to further improve the prediction accuracy and demonstrates the flexibility and adaptability of our SLRC system.
The findings of our study offer a new proof of concept for using physical systems in the context of reservoir computing. Furthermore, they highlight the potential of SL waves as a versatile and powerful tool for complex computations, forecasting, and machine learning tasks. With further research and development, we anticipate that this novel approach could open new avenues in the field of analog computation, especially in situations where conventional digital systems are limited.""
Source: https://www.sciencealert.com/protot...cts-events-better-than-some-digital-computers
"Several theoretical works have shown that solitons---waves that self-maintain constant shape and velocity as they propagate---can be used as a physical computational reservoir, a concept where machine learning algorithms designed for digital computers are replaced by analog physical systems that exhibit nonlinear dynamical behaviour. Here we propose and experimentally validate a novel reservoir computing (RC) system that for the first time employs solitary-like (SL) waves propagating on the surface of a liquid film flowing over an inclined surface. We demonstrate the ability of the SL wave RC system (SLRC) to forecast chaotic time series, also producing experimental evidence of the possibility to combine RC with nonlinear vector autoregression techniques not only in a computer program but also in a physical systems."
---
""Reservoir computing (RC) has evolved significantly over the years as a branch of recurrent neural networks (RNNs) with a uniquely simplified training strategy. The core idea of RC is to take advantage of the inherent dynamics of a high-dimensional system, referred to as the "reservoir," to process temporal information. This paradigm shift has facilitated the exploration of various physical systems as reservoirs, and among them, solitary-like (SL) waves present in liquid film flows over inclined surfaces stand out.
SL waves, a type of soliton wave, maintain their shape and velocity as they travel. They naturally exist in many physical systems, including fluid flows. These waves can be thought of as individual information carriers, each propagating through the system according to its internal dynamics. The hypothesis is that the interaction and evolution of these waves, due to their nonlinear nature, can be harnessed to perform complex computations.
In our study, we designed and validated a SL wave-based RC system (SLRC). We accomplished this by creating a setup where a liquid film flowed over an inclined surface, producing SL waves. We then implemented an input method to perturb the system and a readout mechanism to capture the resulting wave dynamics, thus forming the SLRC.
The most significant highlight of this work is the successful application of SLRC for predicting chaotic time series. Chaotic systems, by definition, are highly sensitive to initial conditions, and their future states can be challenging to predict accurately. However, the nonlinear dynamic behavior exhibited by the SL waves provides a powerful tool to tackle this issue. The results showed that the SLRC could learn and forecast the complex patterns of these systems effectively.
Additionally, we demonstrated the capability of integrating the SLRC with nonlinear vector autoregression techniques. Traditionally, these techniques were used only in digital computation, but here we showed that they could be applied in the physical realm. This integration allows us to further improve the prediction accuracy and demonstrates the flexibility and adaptability of our SLRC system.
The findings of our study offer a new proof of concept for using physical systems in the context of reservoir computing. Furthermore, they highlight the potential of SL waves as a versatile and powerful tool for complex computations, forecasting, and machine learning tasks. With further research and development, we anticipate that this novel approach could open new avenues in the field of analog computation, especially in situations where conventional digital systems are limited.""
Source: https://www.sciencealert.com/protot...cts-events-better-than-some-digital-computers