Electrical engineer Dr. Amir Handelman dabbles in all things AI (artificial intelligence). This highly applicative field of research is currently experiencing exponential growth and is penetrating all aspects of our life—from medical diagnostics to human-like text and image generation and beyond. We met with Dr. Amir for a brief Q&A session about his research.
I have pursued a diverse range of research topics, a direct reflection of my great passion for my field. In the area of medicine, I have developed a computerized model for objectively evaluating cutting-edge performance using a laparoscopic box trainer simulator, and currently I am developing a new sensor for intraoperative monitoring of the corneal tissue during cataract surgery. In the field of optics, I have investigated ways to mitigate the effect of ‘beam wandering’ due to water-waves in underwater wireless optical communication, and implemented diverse machine learning algorithms to gain insight from optical fringe patterns and to extract optical signals transmitted underwater.
I am looking for ways to apply artificial intelligence in optics and biomedicine. It could be AI-based image processing for medical diagnostics, it could be AI in medical ethics, it could be AI as a signal processing approach for medical image analysis, and so on. In optics, it could be gaining more insight into various optical phenomenon. I have already gained headway and published in all these topics.
Photonics deals with the generation, detection and manipulation of light. The applications of photonics, which are numerous, include telecommunications, optical data storage, displays, and light-matter interactions. Currently, I am researching new ways to use different bio-organic materials as the main platform for the fabrication of photonic integrated circuits that can perform advanced computation in hardware.
The idea is to make hardware that can accelerate AI optically by using eco-friendly materials, not inorganic materials as is done today.
The great power of AI is its ability to learn, but it does so via training, which is the most time- and energy-consuming step in AI, so everyone is trying to accelerate it. Increasing the permutation of neural networks and algorithms, as well as reducing the latency and energy consumption as much as possible, requires non-traditional hardware. Current electronic AI hardware, such as those the big tech companies like Nividia and Intel are working on, have limited bandwidth, and they produce a lot of heat. With photonics, we can overcome these two issues. Photonics AI chips can improve performance in terms of both bandwidth and power consumption.
In my current research, we are taking it even further. We are using bio-organic materials as the platform for the fabrication of these photonic integrated circuits, which can perform neuromorphic computations and AI calculations.
Our approach is unique both in terms of fabrication and features. In the fabrication aspect, the uniqueness lies in the introduction of new photonic chips that can potentially allow signal processing operations and can be embedded with other organic-based components. In terms of features, they are flexible and much cheaper, and importantly, they are eco-friendly. Organic AIs are also biocompatible, which means that you can use them for various healthcare applications, such as wearable sensors, biophysical biomarkers, seizure detectors, and so on.
There is no doubt that the future of medicine is in AI. The vision is that everyone will use AI but will also have ready access to AI hardware, like a smartwatch with AI-based sensors. In healthcare, such AI-based sensors could one day govern controlled drug release—in real time. They will do so by measuring hundreds of parameters and learning about your health state and then release drugs when needed—the ultimate expression of personalized medicine. Every two people are different, so you need something adaptive, which is exactly what AI can do.
This vision, however, poses a formidable challenge because AI requires immense processing power. And because of this massive processing requirement, the da typically must be sent to the cloud for processing. But not everyone has constant access to the cloud, and it also overburdens the cloud. So, we are working to make .things local, so that some of the processing is done on the “edge” itself
And there are other ways by which AI can improve medicine. One such topic that I am also working on is to improve medical image processing. I have already initiated collaborations with gastroenterologists in Israel to improve disease diagnostics. But I believe more could be done in this field and would gladly work with physicians around the world or people with access to MRI and CT images who seek to address medical needs using AI.
Today, most people rely on computerized systems in their daily lives, like Waze for navigation, and AI it is also adopted more and more by professionals, such as physicians. But what happens if a medical AI system falters? Who is responsible? Is it the doctor or the developers of the AI program? This is a highly intriguing topic – AI ethics — and I have already started investigating it.
Another example of an ethical issue has to do with a potential hidden agenda of the AI software developers. What happens if a group with a hidden agenda—and this could be anyone (misogynists, racists, anti-Semites, etc.)—working in a certain AI development company are able to codify their agenda into platforms? The outcome could be software that creates bias against a certain group of people, such as who is eligible for early release from prison, whose mortgage is approved, or which applicants are suitable for a certain position. This could lead to hidden discrimination against a particular group. Such implanted bias is extremely difficult to detect. This is very dangerous and needs to be urgently addressed, but it can only be done by experts in the field, which is why I am deeply invested in this issue.
Dr. Amir Handelman joined the Faculty of Electrical Engineering at HIT- Holon Institute of Technology as a tenure-track faculty member after receiving his PhD in Electrical Engineering from Tel-Aviv University, Israel, in 2014. In addition to his academic background, Amir has over 10 years of experience in computer vision and optics, which he gained working in several hi-tech companies, such as Israel Aerospace Industries (IAI), Volume-Elements LTD. and KLA-Tencor. Amir has hands-on experience in developing computer vision and image processing algorithms for industrial applications. In his AI in Optics, Medicine and Materials Lab, his team explores a range of topics: optic sensors, organic photonics, nonlinear optics, free-space optical communication, waveguide structures, AI-photonics, medical instruments, computer vision and machine learning and optical materials. At HIT, Amir teaches courses in image processing, computer vision, optical engineering, optical communication, solid state devices and introduction to electrical engineering.
