Assistant Professor Uğur Teğin from Koç University’s Department of Electrical and Electronics Engineering has developed two new systems that use light itself, rather than conventional electronic circuitry, to perform computations for artificial intelligence. Published in two journals within Nature Portfolio, one of the world’s leading scientific publishing groups, the studies introduce new approaches to computing that could contribute to the development of faster and more energy-efficient AI systems.

As artificial intelligence (AI) technologies become increasingly widespread, the growing demand for processing power and energy is emerging as a global challenge. Two new systems developed at Koç University offer a potential response to this challenge by harnessing the physical properties of light to perform part of the computation required to analyze images.

Research projects led by Assistant Professor Uğur Teğin together with his students, were published in two journals Communications Engineering and Communications Physics within highly respected Nature Portfolio. The two complementary studies explore how light’s ability to carry and process multiple streams of information simultaneously can be harnessed for artificial intelligence applications.

Unlike electronic signals, light can carry and process multiple streams of information simultaneously. This makes photonic computing a powerful alternative for computationally intensive AI applications.

Processing Data with Light Without Losing Color Information

In the first study, published in Communications Engineering, Assistant Professor Uğur Teğin and master’s student Fatma Nur Kılınç developed a programmable photonic neural network that uses light to transform the information in color images into representations that are easier to classify, without discarding their color content.

Many conventional image-processing systems convert color images to grayscale, potentially losing important information contained in the red, green, and blue channels. The system developed at Koç University preserves these channels and processes them simultaneously. Its optical architecture accentuates features such as color, shape, and texture that help distinguish one image from another, enabling more accurate classification.

The data are first encoded onto a programmable device that controls the propagation of light. As the different color channels travel together through the optical system, the device combines and transforms them in a way tailored to the task at hand. Passing through the optical setup accentuates features—such as color, shape, and texture—that help distinguish one image from another. The resulting optical pattern is then recorded by a camera and used to predict the category to which the image belongs.

Adapting to Hardware Imperfections

One of the key advantages of the photonic neural network is its ability to adapt to the characteristics of the physical optical system without requiring a perfectly accurate computer model. In real optical setups, minor misalignments, manufacturing imperfections, and other physical effects can cause experimental results to deviate from idealized simulations. By learning directly from experimental measurements, the photonic neural network can adapt to these non-ideal conditions.

High Classification Accuracy

In a dermatological disease classification task, classification accuracy increased from 68% using raw color pixel data to 98% following optical processing. The researchers also demonstrated that the method could be applied not only to images but to other forms of multidimensional data.

Turning a Fiber Laser into a Physical AI Processor

In the second study, published in Communications Physics, Assistant Professor Uğur Teğin, master’s students Dilem Eşlik and Fatma Nur Kılınç, and undergraduate student Bahadır Utku Kesgin used a multimode fiber laser as a physical processor for machine learning.
Information from images was encoded into light and processed within the laser cavity, transforming it into optical patterns that were easier to classify. In this way, the physical dynamics of the laser itself performed part of the computation that would ordinarily be carried out by computer software.

Strong Performance Across Diverse Datasets

The system was tested using images of recyclable waste, weather and satellite imagery, retinal scans, and skin lesions. Following optical processing, classification accuracy ranging from 85 to 99 percent were achieved across the different datasets using only a simple classification method. Because the laser itself performed a substantial share of the computation, the system also required far fewer trainable parameters than conventional deep-learning models.

Toward More Energy-Efficient and Sustainable AI

When viewed in tandem, the two studies conducted at Koç University demonstrate complementary strengths of photonic computing. The first shows how light can process multidimensional information while preserving valuable features such as color, while the second demonstrates how the physical dynamics of a laser can take on part of the computational workload itself.

Although both systems still rely on electronic components such as cameras and control units, advances in high-speed light-control devices and compact, chip-integrated optical systems could enable light to assume a much greater share of the workload currently handled by electronic processors.
By shifting more computational operations from electronics to light, these technologies could help pave the way for a new generation of faster, more energy-efficient and sustainable artificial intelligence systems.

The Publications:

“Self-optimizing multichannel optical computing”, Communications Engineering, 2026.
“Multimode fiber laser cavities as nonlinear optical processors”, Communications Physics, 2026.

About Assistant Proffesor Uğur Teğin:
Asst. Prof. Uğur Teğin has been a faculty member in the Department of Electrical and Electronics Engineering at Koç University since 2023. After completing his bachelor’s degree in physics and his master’s degree in Materials Science and Nanotechnology at Bilkent University, Teğin earned his PhD in Photonics from the Swiss Federal Institute of Technology Lausanne (EPFL). Subsequently, he conducted postdoctoral research at the California Institute of Technology (Caltech). His research focuses on ultrafast lasers, optical computing, and photonic neural networks.

About Koc University:
Established in 1993 with the mission of cultivating highly competent graduates, advancing the frontiers of science, and serving the country and humanity, Koç University is an institution providing world-class education. Offering 22 undergraduate, 44 master’s, and 30 doctoral programs, the university has a student body of over 9,000, with 65 percent of students studying on scholarships. To date, more than 20,000 students have graduated from Koç University’s undergraduate and graduate programs. With its extensive, world-class laboratory, computing, and research facilities, Koç University ranks among the top educational institutions in Turkey in terms of the number of scientific articles published per faculty member.