Fundamental Research
Deep Learning
With access to a vast amount of data, we focus on modeling the underlying composition of both structured and unstructured data. Our research involves fundamental and theoretical work in the field of deep learning, including mathematics, statistics and algorithms. The most critical mission of IIAI is to advance our understanding of the underlying mechanisms giving rise to intelligence. As such, we dedicate substantial efforts to exploring high-level reasoning and thinking skills, with the goal of creating human-imitative intelligence. Any advances in the development of fundamental machine learning algorithms and their application to real world scenarios will accelerate innovation and result in a significant impact on society. Our primary research areas include, but are not limited to: deep neural network understanding, unsupervised learning, and adversarial learning.
Reinforcement Learning
Reinforcement Learning (RL) is an important area of machine learning, inspired by behaviorist psychology. With RL, an AI agent learns to behave in a given environment by performing actions that maximize a cumulative reward. The agent operates by getting feedback information from their own actions and experiences, learning through trial and error. RL is a strong focus at IIAI, where research is conducted into both policy-search and temporal-difference methods, as well as other closely related methods for sample-based decision theoretic planning. Efforts will be dedicated to developing AI agents that are able to learn in interactive environments by exploring solutions and automatically determining the ideal behavior. A variety of techniques are expected to play key roles in this research, including Monte Carlo tree search, swarm intelligence, control theory and deep learning, amongst others. Finally, we aim to apply RL to real-word scenarios, such as automatic natural language processing, speech analysis, healthcare, image processing and video analytics.
Computer Vision
Computer vision is an interdisciplinary field that seeks to build a human-imitative visual-understanding system that enables machines to gather high-level understanding from multiple visual data forms (e.g. images and video sequences). At IIAI we have two goals in this area. Firstly, from a scientific research perspective, we aim to further develop the computer vision theories behind the artificial visual understanding system. With this, we aim to solve a wide range of fundamental computer vision problems, including human action recognition, gait recognition, and person re-identification, amongst others. From a technological perspective, we aim to then transfer our research findings to enhancing the performance of already existing visual-understanding systems.
Natural Language Processing
Language is an advanced and complex system of human communication. Natural Language Processing (NLP) is an area of research that explores the capabilities of machines to understand and make use of human languages. Based on the latest advancement in NLP theories, IIAI aims to develop innovative text input and understanding solutions to create a universal and flexible interface, enabling machines to intelligently and accurately understand voices and texts. In addition to this, IIAI also aims to develop highly effective and efficient NLP techniques capable of dealing with large volumes of natural language data within multiple forms of documentation. Eventually, we aim to build an artificial language system that will be the interface not only between machines and humans, but also between the whole textual and linguistic world.
Applied Research
Video Understanding
We aim to build a world-leading, video-understanding platform that could unlock a new era of public and private monitoring devices. This would involve the application of highly sophisticated video analytics to improve face detection and annotation, pedestrian identification and tracking, scene analysis and tagging, human behavior analysis, and content enhancement. For example, with access to a national database storing numerous photo IDs, deep-learning techniques could be applied to a nationwide video system to develop a fully-integrated person-detection, tracking and recognition system with the ability to rapidly identify and track a target. Even more mission-critically, we could go from “unknown” to “unknown,” developing a self-learning AI system for the detection of unknown human behavior from real-time surveillance camera feed, which could identify, classify and report abnormal activities in unlimited behavioral forms. Self-taught learning algorithms could study abnormal human patterns across the board and apply inferences to future patterns yet unknown.
Medical Imaging
Medical image computing is an interdisciplinary field at the intersection of computer science, data science, and mathematics, with a wide range of applications to medicine and bioscience. At IIAI, researchers are fully dedicated to applying cutting-edge machine learning and computer vision techniques to revolutionize the healthcare industry through medical image computation and analysis. This involves developing computer-assisted interventional systems and robotics, computer-aided diagnosis systems, and clinical visualization systems, amongst others. All these applications require the manipulation and integration of medical image information. Many applications also depend on the integration of image information with sensor data (e.g. from tracking systems), effector control systems (e.g. robots or positioning devices), visual displays, or other feedback systems. As such, our researchers focus on a broad range of medical imaging applications. For instance, one of our key applications will be destined to mammography, which is a medical imaging technique that uses a low-dose X-ray system to aid in the early detection and diagnosis of breast diseases in women. Advanced deep learning techniques can be applied to detect cancerous lesions much more accurately and efficiently than the traditional examinations by expert radiologists. Another challenging target is Gastrointestinal Infections (GI), where AI-based medical imaging can be deployed to attain highly effective detection and classification. Finally, medical image computing techniques are also used in eye fundus scans for the early detection of diabetes and cardiac disease diagnosis.
Healthcare
AI has numerous applications to healthcare. At IIAI, we aim to redesign healthcare by using the latest advancement in AI research to target drug development, computer-assisted interventional systems and robotics, bioscience applications and computer-aided diagnosis. One important area of focus is personalized healthcare. AI has many applications to this area. For example, AI algorithms have been developed to help diabetic patients control their glucose levels by gathering data on how food, exercise, and insulin affect their glucose levels under different conditions. The advice stemming from these algorithms could help more than 30 million Americans suffering from diabetes. Additionally, a care manager might use an office-based platform to obtain a trend analysis on the peaks and dips in an individual’s blood glucose levels during certain days, weeks, or months. The results could be used to adjust a patient’s overall care plan.