Amin Shafiee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Amin Shafiee has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Mahdi Nikdast
Colorado State University, Fort Collins, CO, USA
15 shared publications - Sudeep Pasricha
Colorado State University
9 shared publications - Febin Sunny
Department of Electrical and Computer Engineering Colorado State University Fort Collins CO USA
4 shared publications - Benoit Charbonnier
Universit´e Grenoble Alpes, CEA-Leti
3 shared publications - Zahra Ghanaatian
Department of Electrical and Computer Engineering, Colorado State University, Fort Collins, CO, USA
3 shared publications - Asif Mirza
Department of Electrical and Computer Engineering, Colorado State University, Fort Collins, CO, USA
2 shared publications
Latest Publications
- 2 July 2026
- Published by Springer Nature
- p. 201-237
Accelerating AI Applications with Silicon Photonics: Status and Perspective
- 2 July 2026
- Published by Springer Nature
- p. 201-237
Accelerating AI Applications with Silicon Photonics: Status and Perspective
- Open Access
- 17 June 2026
- Published by Springer Nature in Communications Engineering
- Vol. 5 (1)
LightPro: a linear photonic processor with full programmability
The physical scaling of photonic matrix-vector multiplication hardware for deep neural network acceleration is fundamentally limited by accumulated optical losses, crosstalk noise, and the prohibitive footprint of conventional devices such as Mach-Zehnder interferometers. Here we present LightPro, a fully programmable linear photonic processor designed to optimize scalability, power efficiency, and area footprint. At its core, our architecture integrates a neural architecture search and pruning framework with tunable phase-change material directional couplers. By thermally modulating the phase-change material state, we dynamically adjust coupling coefficients to achieve precise splitting ratios, facilitating highly optimized topologies for matrix-vector multiplication operations. The underlying phase-change material-based devices are evaluated using numerical multiphysics simulations and compact models, which are validated against reported experimental data from prior work. System-level evaluations demonstrate that the neural architecture search-optimized LightPro architectures achieve up to an 85% footprint reduction and a greater than 50% decrease in power consumption. Network scaling evaluations using handwritten digit and Gaussian datasets yield an inference accuracy degradation of less than 5%. Experimental prototyping on a commercial photonic processor validates the computational accuracy of LightPro, establishing a scalable and efficient pathway for next-generation photonic artificial intelligence accelerators.
- 17 June 2026
- Published by Springer Nature in Communications Engineering
- Vol. 5 (1)
Open AccessLightPro: a linear photonic processor with full programmability
The physical scaling of photonic matrix-vector multiplication hardware for deep neural network acceleration is fundamentally limited by accumulated optical losses, crosstalk noise, and the prohibitive footprint of conventional devices such as Mach-Zehnder interferometers. Here we present LightPro, a fully programmable linear photonic processor designed to optimize scalability, power efficiency, and area footprint. At its core, our architecture integrates a neural architecture search and pruning framework with tunable phase-change material directional couplers. By thermally modulating the phase-change material state, we dynamically adjust coupling coefficients to achieve precise splitting ratios, facilitating highly optimized topologies for matrix-vector multiplication operations. The underlying phase-change material-based devices are evaluated using numerical multiphysics simulations and compact models, which are validated against reported experimental data from prior work. System-level evaluations demonstrate that the neural architecture search-optimized LightPro architectures achieve up to an 85% footprint reduction and a greater than 50% decrease in power consumption. Network scaling evaluations using handwritten digit and Gaussian datasets yield an inference accuracy degradation of less than 5%. Experimental prototyping on a commercial photonic processor validates the computational accuracy of LightPro, establishing a scalable and efficient pathway for next-generation photonic artificial intelligence accelerators.
- Open Access
- 13 November 2025
- Published by Springer Nature in Research Square
LightPro: A Linear Photonic Processor with Full Programmability
Silicon photonics (SiPh) enables integration of optical systems for applications including high-speed data communication, optical I/O, and energy-efficient optical computing for AI accelerators. Photonic AI acceleration aims to improve matrix-vector multiplication (MVM), the most energy-intensive computation in deep neural networks (DNNs). As DNNs grow in complexity, larger photonic MVM networks are required, leading to accumulated optical losses, phase and crosstalk noise, and large device footprints (e.g., Mach–Zehnder Interferometers), limiting scalability and efficiency. This paper proposes a fully programmable linear photonic processor, LightPro, with improved scalability, performance, and footprint. LightPro uses compact, low-loss, and programmable SiPh directional coupler (DC) devices employing phase-change material (PCM) to program the DC’s splitting ratio by thermally inducing phase transitions. Based on this device, a neural architecture search (NAS) and pruning algorithm optimizes the processor for MVM operations. Simulation results show up to 85% footprint reduction and over 50% improvement in power consumption. LightPro performs inference with weight matrices trained on MNIST and Gaussian datasets, showing less than 5% drop in accuracy when scaling. Prototyping with the iPronics SmartLight processor demonstrates LightPro’s efficiency, computational accuracy, and feasibility for next-generation photonic AI accelerators
- 13 November 2025
- Published by Springer Nature in Research Square
Open AccessLightPro: A Linear Photonic Processor with Full Programmability
Silicon photonics (SiPh) enables integration of optical systems for applications including high-speed data communication, optical I/O, and energy-efficient optical computing for AI accelerators. Photonic AI acceleration aims to improve matrix-vector multiplication (MVM), the most energy-intensive computation in deep neural networks (DNNs). As DNNs grow in complexity, larger photonic MVM networks are required, leading to accumulated optical losses, phase and crosstalk noise, and large device footprints (e.g., Mach–Zehnder Interferometers), limiting scalability and efficiency. This paper proposes a fully programmable linear photonic processor, LightPro, with improved scalability, performance, and footprint. LightPro uses compact, low-loss, and programmable SiPh directional coupler (DC) devices employing phase-change material (PCM) to program the DC’s splitting ratio by thermally inducing phase transitions. Based on this device, a neural architecture search (NAS) and pruning algorithm optimizes the processor for MVM operations. Simulation results show up to 85% footprint reduction and over 50% improvement in power consumption. LightPro performs inference with weight matrices trained on MNIST and Gaussian datasets, showing less than 5% drop in accuracy when scaling. Prototyping with the iPronics SmartLight processor demonstrates LightPro’s efficiency, computational accuracy, and feasibility for next-generation photonic AI accelerators
- Open Access
LightPro: A Linear Photonic Processor with Full Programmability
In this paper, we propose a novel fully programmable linear photonic processor, which we call LightPro, with improved scalability, performance, and footprint. At the heart of LightPro are compact, low-loss, and programmable silicon photonic (SiPh) directional coupler (DC) devices that deploy phase-change material (PCM) for programming the DC's splitting ratio. By thermally inducing phase transitions in the PCM, the coupling coefficient of the DC can be dynamically adjusted to achieve different splitting ratios in the device output. Building on this device foundation, we develop a neural architecture search (NAS) and pruning algorithm to optimize the architecture of the processor for performing MVM operations. Our simulation results show that LightPro achieves up to an 85% reduction in footprint and more than 50% improvement in power consumption. In addition, LightPro is evaluated by performing inference with weight matrices trained on MNIST and linearly separable Gaussian datasets, showing less than a 5% drop in accuracy when scaling up the network. Prototyping results, using a commercial photonic processor (iPronics SmartLight), show LightPro's efficiency and performance (e.g., computational accuracy) compared to conventional photonic MVM hardware, demonstrating the experimental evaluation and feasibility of LightPro for next-generation photonic AI accelerators.
Open AccessLightPro: A Linear Photonic Processor with Full Programmability
In this paper, we propose a novel fully programmable linear photonic processor, which we call LightPro, with improved scalability, performance, and footprint. At the heart of LightPro are compact, low-loss, and programmable silicon photonic (SiPh) directional coupler (DC) devices that deploy phase-change material (PCM) for programming the DC's splitting ratio. By thermally inducing phase transitions in the PCM, the coupling coefficient of the DC can be dynamically adjusted to achieve different splitting ratios in the device output. Building on this device foundation, we develop a neural architecture search (NAS) and pruning algorithm to optimize the architecture of the processor for performing MVM operations. Our simulation results show that LightPro achieves up to an 85% reduction in footprint and more than 50% improvement in power consumption. In addition, LightPro is evaluated by performing inference with weight matrices trained on MNIST and linearly separable Gaussian datasets, showing less than a 5% drop in accuracy when scaling up the network. Prototyping results, using a commercial photonic processor (iPronics SmartLight), show LightPro's efficiency and performance (e.g., computational accuracy) compared to conventional photonic MVM hardware, demonstrating the experimental evaluation and feasibility of LightPro for next-generation photonic AI accelerators.
- 30 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Emerging and Selected Topics in Circuits and Systems
- Vol. 16 (1) , 52-66
Robust-by-Design Silicon Photonic Mach–Zehnder Interferometers Under Fabrication Nonuniformities
Silicon photonics (SiPh) has quickly emerged as a leading technology for photonic integrated circuits, offering advantages such as low latency, low power consumption, and high bandwidth in different application domains from data communication to computation. Among different SiPh devices, Mach–Zehnder interferometers (MZIs) are highly attractive for applications such as modulators, switches, biosensors, and neural networks. However, MZIs are sensitive to fabrication-process variations (FPVs), which can significantly impact MZI performance even with minute changes in the device’s critical dimensions. To address this problem, we present a variation-aware design optimization method for MZIs to minimize optical response shift and changes in free spectral range (FSR), and maximize extinction ratio (ER) under FPVs. We develop detailed analytical models to accurately capture the impact of FPVs in MZIs. Leveraging these models, device layout and design parameters, such as waveguide width, are efficiently explored and optimized based on single-mode and multi-mode design considerations to improve device resilience under FPVs. The simulation results show that, compared to conventional MZIs, our optimized MZIs result in more than a 57% reduction in optical-response shift, as well as FSR and ER changes under FPVs. Moreover, the simulation results show great agreement with the measured optical response of fabricated MZIs. We also present a system-level case study of an MZI-based photonic neural network under FPVs, showcasing up to a 65% improvement in the network inferencing accuracy when using optimized MZIs.
- 30 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Emerging and Selected Topics in Circuits and Systems
- Vol. 16 (1) , 52-66
Robust-by-Design Silicon Photonic Mach–Zehnder Interferometers Under Fabrication Nonuniformities
Silicon photonics (SiPh) has quickly emerged as a leading technology for photonic integrated circuits, offering advantages such as low latency, low power consumption, and high bandwidth in different application domains from data communication to computation. Among different SiPh devices, Mach–Zehnder interferometers (MZIs) are highly attractive for applications such as modulators, switches, biosensors, and neural networks. However, MZIs are sensitive to fabrication-process variations (FPVs), which can significantly impact MZI performance even with minute changes in the device’s critical dimensions. To address this problem, we present a variation-aware design optimization method for MZIs to minimize optical response shift and changes in free spectral range (FSR), and maximize extinction ratio (ER) under FPVs. We develop detailed analytical models to accurately capture the impact of FPVs in MZIs. Leveraging these models, device layout and design parameters, such as waveguide width, are efficiently explored and optimized based on single-mode and multi-mode design considerations to improve device resilience under FPVs. The simulation results show that, compared to conventional MZIs, our optimized MZIs result in more than a 57% reduction in optical-response shift, as well as FSR and ER changes under FPVs. Moreover, the simulation results show great agreement with the measured optical response of fabricated MZIs. We also present a system-level case study of an MZI-based photonic neural network under FPVs, showcasing up to a 65% improvement in the network inferencing accuracy when using optimized MZIs.
- Open Access
- 29 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Event-Driven Spatiotemporal Processing-In-Sensor with Phase Change Memory-based Optical Acceleration
- 29 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessEvent-Driven Spatiotemporal Processing-In-Sensor with Phase Change Memory-based Optical Acceleration
- Open Access
- 20 January 2025
- Published by Association for Computing Machinery (ACM)
- p. 128-134
Invited paper: Bridging EDA and Silicon Photonics Design: Enabling Robust-by-Design Photonic Integrated Circuits
Silicon photonic devices are essential components of integrated optical communication systems and emerging photonic processors. However, their performance is notably impacted by fabrication-process variations (FPVs), which primarily stem from optical lithography imperfections. The impact of FPVs can accumulate and deteriorate the system-level performance through, for example, increasing system power consumption, accumulated crosstalk noise, and degrading signal integrity in photonic systems. In this paper, we discuss the promise of variation-aware design-space exploration and optimization to enhance photonic device robustness under different FPVs while considering two silicon photonic devices used widely in different applications, namely Microring Resonators (MRRs) and Mach-Zehnder Interferometers (MZIs). In addition, we consider a system-level case study of an MZI-based coherent neural network, where we show how our proposed variation-aware design optimization at the device level helps improve the network accuracy by up to 88% under FPVs.
- 20 January 2025
- Published by Association for Computing Machinery (ACM)
- p. 128-134
Open AccessInvited paper: Bridging EDA and Silicon Photonics Design: Enabling Robust-by-Design Photonic Integrated Circuits
Silicon photonic devices are essential components of integrated optical communication systems and emerging photonic processors. However, their performance is notably impacted by fabrication-process variations (FPVs), which primarily stem from optical lithography imperfections. The impact of FPVs can accumulate and deteriorate the system-level performance through, for example, increasing system power consumption, accumulated crosstalk noise, and degrading signal integrity in photonic systems. In this paper, we discuss the promise of variation-aware design-space exploration and optimization to enhance photonic device robustness under different FPVs while considering two silicon photonic devices used widely in different applications, namely Microring Resonators (MRRs) and Mach-Zehnder Interferometers (MZIs). In addition, we consider a system-level case study of an MZI-based coherent neural network, where we show how our proposed variation-aware design optimization at the device level helps improve the network accuracy by up to 88% under FPVs.
- 6 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 43 (11) , 3888-3899
OPIMA: Optical Processing-in-Memory for Convolutional Neural Network Acceleration
Recent advances in machine learning (ML) have spotlighted the pressing need for computing architectures that bridge the gap between memory bandwidth and processing power. The advent of deep neural networks has pushed traditional Von Neumann architectures to their limits due to the high latency and energy consumption costs associated with data movement between the processor and memory for these workloads. One of the solutions to overcome this bottleneck is to perform computation within the main memory through processing-in-memory (PIM), thereby limiting data movement and the costs associated with it. However, dynamic random-access memory-based PIM struggles to achieve high throughput and energy efficiency due to internal data movement bottlenecks and the need for frequent refresh operations. In this work, we introduce OPIMA, a PIM-based ML accelerator, architected within an optical main memory. OPIMA has been designed to leverage the inherent massive parallelism within main memory while performing high-speed, low-energy optical computation to accelerate ML models based on convolutional neural networks. We present a comprehensive analysis of OPIMA to guide design choices and operational mechanisms. In addition, we evaluate the performance and energy consumption of OPIMA, comparing it with conventional electronic computing systems and emerging photonic PIM architectures. The experimental results show that OPIMA can achieve$2.98\times $ higher throughput and$137\times $ better energy efficiency than the best known prior work.- 6 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 43 (11) , 3888-3899
OPIMA: Optical Processing-in-Memory for Convolutional Neural Network Acceleration
Recent advances in machine learning (ML) have spotlighted the pressing need for computing architectures that bridge the gap between memory bandwidth and processing power. The advent of deep neural networks has pushed traditional Von Neumann architectures to their limits due to the high latency and energy consumption costs associated with data movement between the processor and memory for these workloads. One of the solutions to overcome this bottleneck is to perform computation within the main memory through processing-in-memory (PIM), thereby limiting data movement and the costs associated with it. However, dynamic random-access memory-based PIM struggles to achieve high throughput and energy efficiency due to internal data movement bottlenecks and the need for frequent refresh operations. In this work, we introduce OPIMA, a PIM-based ML accelerator, architected within an optical main memory. OPIMA has been designed to leverage the inherent massive parallelism within main memory while performing high-speed, low-energy optical computation to accelerate ML models based on convolutional neural networks. We present a comprehensive analysis of OPIMA to guide design choices and operational mechanisms. In addition, we evaluate the performance and energy consumption of OPIMA, comparing it with conventional electronic computing systems and emerging photonic PIM architectures. The experimental results show that OPIMA can achieve$2.98\times $ higher throughput and$137\times $ better energy efficiency than the best known prior work.- Open Access
- 1 July 2024
- Published by SPIE-Intl Soc Optical Eng in Journal of Optical Microsystems
- Vol. 4 (03) , 031208-19
Programmable phase change materials and silicon photonics co-integration for photonic memory applications: a systematic study
In the post-Moore era, conventional CMOS-based integrated circuits are struggling to meet the performance requirements related to emerging data-driven applications, which include large language models, intrusion detection systems, and pandemic growth detection systems. In particular, conventional electronic memory technologies, such as DRAMs and SRAMs, are facing constraints due to manufacturing scalability concerns, data retention problems, power penalties, leakage issues, and bandwidth limitations.1 – 3 Nonvolatile memories (NVMs) based on ferroelectric materials, such as FeRAMs, and resistive metal oxides, such as ReRAMs, can address the data retention issues related to DRAMs and SRAMs. Despite the promise of NVMs, they suffer from reliability issues and limited write endurance. This motivates the need for a new alternative NVM for emerging communication and computation systems.
- 1 July 2024
- Published by SPIE-Intl Soc Optical Eng in Journal of Optical Microsystems
- Vol. 4 (03) , 031208-19
Open AccessProgrammable phase change materials and silicon photonics co-integration for photonic memory applications: a systematic study
In the post-Moore era, conventional CMOS-based integrated circuits are struggling to meet the performance requirements related to emerging data-driven applications, which include large language models, intrusion detection systems, and pandemic growth detection systems. In particular, conventional electronic memory technologies, such as DRAMs and SRAMs, are facing constraints due to manufacturing scalability concerns, data retention problems, power penalties, leakage issues, and bandwidth limitations.1 – 3 Nonvolatile memories (NVMs) based on ferroelectric materials, such as FeRAMs, and resistive metal oxides, such as ReRAMs, can address the data retention issues related to DRAMs and SRAMs. Despite the promise of NVMs, they suffer from reliability issues and limited write endurance. This motivates the need for a new alternative NVM for emerging communication and computation systems.
- 5 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (13) , 4598-4613
Analysis of Optical Loss and Crosstalk Noise in MZI-Based Coherent Photonic Neural Networks
With the continuous increase in the size and complexity of machine learning models, the need for specialized hardware to efficiently run such models is rapidly growing. To address such a need, silicon-photonic-based neural network (SP-NN) accelerators have recently emerged as a promising alternative to electronic accelerators due to their lower latency and higher energy efficiency. Not only can SP-NNs alleviate the fan-in and fan-out problem with linear algebra processors, their operational bandwidth can match that of the photodetection rate (typically $\approx$ 100 GHz), which is at least over an order of magnitude faster than electronic counterparts that are restricted to a clock rate of a few GHz. Unfortunately, the underlying silicon photonic devices in SP-NNs suffer from inherent optical losses and crosstalk noise originating from fabrication imperfections and undesired optical couplings, the impact of which accumulates as the network scales up. Consequently, the inferencing accuracy in an SP-NN can be affected by such inefficiencies—e.g., can drop to below 10%—the impact of which is yet to be fully studied. In this paper, we comprehensively model the optical loss and crosstalk noise using a bottom-up approach, from the device to the system level, in coherent SP-NNs built using Mach–Zehnder interferometer (MZI) devices. The proposed models can be applied to any SP-NN architecture with different configurations to analyze the effect of loss and crosstalk. Such an analysis is important where there are inferencing accuracy and scalability requirements to meet when designing an SP-NN. Using the proposed analytical framework, we show a catastrophic inferencing accuracy drop of up to 84% for SP-NNs of different scales with three known MZI mesh configurations (i.e., Reck, Clements, and Diamond) due to accumulated optical loss and crosstalk noise.
- 5 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (13) , 4598-4613
Analysis of Optical Loss and Crosstalk Noise in MZI-Based Coherent Photonic Neural Networks
With the continuous increase in the size and complexity of machine learning models, the need for specialized hardware to efficiently run such models is rapidly growing. To address such a need, silicon-photonic-based neural network (SP-NN) accelerators have recently emerged as a promising alternative to electronic accelerators due to their lower latency and higher energy efficiency. Not only can SP-NNs alleviate the fan-in and fan-out problem with linear algebra processors, their operational bandwidth can match that of the photodetection rate (typically $\approx$ 100 GHz), which is at least over an order of magnitude faster than electronic counterparts that are restricted to a clock rate of a few GHz. Unfortunately, the underlying silicon photonic devices in SP-NNs suffer from inherent optical losses and crosstalk noise originating from fabrication imperfections and undesired optical couplings, the impact of which accumulates as the network scales up. Consequently, the inferencing accuracy in an SP-NN can be affected by such inefficiencies—e.g., can drop to below 10%—the impact of which is yet to be fully studied. In this paper, we comprehensively model the optical loss and crosstalk noise using a bottom-up approach, from the device to the system level, in coherent SP-NNs built using Mach–Zehnder interferometer (MZI) devices. The proposed models can be applied to any SP-NN architecture with different configurations to analyze the effect of loss and crosstalk. Such an analysis is important where there are inferencing accuracy and scalability requirements to meet when designing an SP-NN. Using the proposed analytical framework, we show a catastrophic inferencing accuracy drop of up to 84% for SP-NNs of different scales with three known MZI mesh configurations (i.e., Reck, Clements, and Diamond) due to accumulated optical loss and crosstalk noise.