Time-Delayed Reservoir Computing Based on a Two-Element Phased Laser Array for Image Identification

Huang, Yu; Zhou, Pei; Yang, Yigong; Chen, Taiyi; Li, Nianqiang

doi:10.1109/jphot.2021.3115598

Cited by 23 publications

(12 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For BW = R the best performance was 95.95% at 50 Gps with an equivalent processing rate. In other state-of-the-art schemes accuracy similar performance was achieved but with more complicated architectures [44,45], whereas higher performance also requires extra pre-processing of the data and very complicated structures with thousands of neurons [46]. In terms of processing rate, to the best of our knowledge the proposed FP-TDELM surpasses previous implementations relying RC inspired techniques due to the fast and parallel insertion of data achieving a processing time of 3.92 ns per image whereas other works require more than 17.1 ns for the same task [45,46].…”

Section: Fp Tdelm and Sptm For Image Classificationmentioning

confidence: 73%

Multimode Fabry-Perot laser as a reservoir computing and extreme learning machine photonic accelerator

Skontranis,

Sarantoglou,

Sozos

et al. 2023

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

In this work, we introduce Fabry-Perot lasers as neuromoprhic nodes in the context of time-delayed reservoir computing and extreme learning machine for the processing of temporal signals and the high-speed classification of images. By exploiting the multi-wavelength emission capabilities of the Fabry – Perot lasers, additional processing nodes can be introduced, thus raising the computational power without sacrificing processing speed. An experimental validation of this concept using a Fabry-Perot extreme learning machine is presented targeting a time depedent task such as channel equalization for a 50 km 28 Gbaud PAM-4 transmission, offering HD-FEC compatible performance. Additionally, the Fabry-Perot neuromorphic concept has been further strengthened by modyfying the data entry technique by parallelelly assigning different samples of the input signal to different modes so as to significantly reduce speed penalty. Numerical simulations revealed that this alternative data insertion technique can offer a reduction of the processing delay and physical footpritnt by 75 % compared to the the conventional approach assigning the same symbols to all Fabry-Perot modes. Moreover, by using a similar data processing scheme in MNIST image classification task we were able to numerically achieve a processing speed of 255.1 Mimages/sec and a classification accuracy up to 95.95 %.

show abstract

Section: Fp Tdelm and Sptm For Image Classificationmentioning

confidence: 73%

Multimode Fabry-Perot laser as a reservoir computing and extreme learning machine photonic accelerator

Skontranis,

Sarantoglou,

Sozos

et al. 2023

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

show abstract

“…We analyzed the timing information of the spike sequence through an ordinal timeseries analysis [22]. Firstly, a pulse sequence with a length of N is divided into N-D vectors of length D, as shown in Equation (5). ∆t(i) is the time interval of pulse i, and t(i) is the time at which pulse i occurs.…”

Section: Toolsmentioning

confidence: 99%

“…Semiconductor lasers with optical feedback can generate various nonlinear physical phenomena and can be applied in secure communication, random number generation, lidar, comprehensive sensing, and reservoir computing, etc. [1][2][3][4][5][6][7]. Moreover, the low frequency fluctuation (LFF) dynamic induced by the optical feedback architecture has been found to be a type of excitable behavior [8], which can be exploited for photonic neurons [9][10][11][12][13] and neuromorphic computations such as pattern recognition, logic operations, and calculations [14][15][16][17][18][19].…”

Section: Introductionmentioning

confidence: 99%

Numerical Demonstration of the Transmission of Low Frequency Fluctuation Dynamics Generated by a Semiconductor Laser with Optical Feedback

Dou

Qiu

2022

Photonics

View full text Add to dashboard Cite

In this paper, the transmission mechanism of the spike information embedded in the low frequency fluctuation (LFF) dynamic in a cascaded laser system is numerically demonstrated. In the cascaded laser system, the LFF waveform is first generated by a drive laser with optical feedback and is then injected into a response laser. The range of crucial system parameters that can make the response laser generate the LFF dynamic is studied, and the effect of parameter mismatch on the transmission of LFF dynamics is explored through a method of symbolic time-series analysis and the index, such as the spike rate and the cross-correlation coefficient. The results show that the mismatch of the pump current has a more significant influence on the transmission of LFF waveforms than that of the internal physical parameter of the laser, such as the linewidth enhancement factor. Moreover, increasing the injection strength can enhance the robustness of LFF transmission. As spikes of the LFF dynamic generated by lasers with optical feedback is similar to the spike of neurons, the results of this paper can help understanding the information transporting and processing inside the photonic neurons.

show abstract

“…There are mainly two approaches to overcoming the problem. One is creating a spatiotemporal photonic reservoir stemming from the free-space optical and diffractive optical element. , The time-delayed P-RC consisting of the nonlinear node with the delay feedback loop shows great potential to minimize complexity. − The size of the reservoir could be adjusted easily by changing sampled intervals or the feedback time. In 2012, Larger et al proposed the first optoelectronic delay RC system based on a Mach–Zehnder modulator (MZM).…”

Section: Introductionmentioning

confidence: 99%

Photonic Reservoir Computing System for Pattern Recognition Based on an Array of Four Distributed Feedback Lasers

Guo,

Zhou,

Xiang

et al. 2024

ACS Photonics

View full text Add to dashboard Cite

Reservoir computing (RC), especially photonic RC based on a single semiconductor laser with a feedback loop, as a method of machine learning, shows excellent performance in time series prediction and classification tasks. The faster the processing speed, the shorter the feedback loop should be employed. However, the performance of the system is not ideal enough due to the limited reservoir nodes caused by the short feedback loop. To overcome this drawback, it is necessary and challenging to develop integrated photonic RC. In this paper, we propose an integrated neuromorphic photonic time-delayed RC scheme based on an array of four distributed feedback lasers (F-DFBs) with optical feedback and injection. Here, we investigate the feasibility of using larger laser arrays and shorter external feedback cavities to provide the RC system with more virtual nodes at the same processing speed, enabling it to handle more complex tasks. Additionally, we compare the performance of the systems with a single DFB (the S-DFB RC) and the F-DFBs RC through simulations and experiments involving iris recognition tasks. Furthermore, the minimum symbol error rate values can reach 0.052 in experiments (and 0 in simulations).

show abstract

Time-Delayed Reservoir Computing Based on a Two-Element Phased Laser Array for Image Identification

Cited by 23 publications

References 51 publications

Multimode Fabry-Perot laser as a reservoir computing and extreme learning machine photonic accelerator

Multimode Fabry-Perot laser as a reservoir computing and extreme learning machine photonic accelerator

Numerical Demonstration of the Transmission of Low Frequency Fluctuation Dynamics Generated by a Semiconductor Laser with Optical Feedback

Photonic Reservoir Computing System for Pattern Recognition Based on an Array of Four Distributed Feedback Lasers

Contact Info

Product

Resources

About