Shozo Yokoyama, a biology professor at Emory University specializing in color vision evolution, was interviewed by KAUST. Yokoyama's lab identified amino acids regulating red-green and UV vision in vertebrates. He emphasizes the importance of young scientists developing fresh perspectives on evolution and learning directly from animals. Why it matters: While not directly an AI story, the piece highlights KAUST's broader research focus and its investment in attracting and showcasing international scientific expertise, relevant to building a strong research ecosystem.
Margaret Livingstone, a neurobiology professor at Harvard Medical School, lectured at KAUST's Winter Enrichment Program 2018 on how art can reveal insights into the human brain. She discussed how artists have long understood the independent roles of color and luminance in visual perception. Livingstone highlighted examples from Picasso, Monet, and Warhol to illustrate how artists manipulate visual cues. Why it matters: This interdisciplinary approach can potentially lead to new understandings of how the brain processes visual information and inform advances in both neuroscience and art.
MBZUAI Professor Fahad Khan is working on a unified theory of machine visual intelligence. His goal is to enable AI systems to better understand and function in complex, chaotic visual environments. The aim is to improve real-world applications like smart cities, personalized healthcare, and autonomous vehicles. Why it matters: This research could significantly advance AI's ability to perceive and interact with the real world, especially in challenging environments common in the developing world.
Dr. Andrew Bastawrous, CEO/co-founder of Peek, discussed his work on mobile eye clinics at KAUST. He developed Peek Acuity and Peek Retina, which turn smartphones into tools for detecting visual impairment. The technology uses smartphone screens and camera clip-ons to image inside the eye. Why it matters: This low-cost mobile ophthalmic tool has the potential to prevent and treat vision loss in underserved communities.
The paper introduces the Prism Hypothesis, which posits a correspondence between an encoder's feature spectrum and its functional role, with semantic encoders capturing low-frequency components and pixel encoders retaining high-frequency information. Based on this, the authors propose Unified Autoencoding (UAE), a model that harmonizes semantic structure and pixel details using a frequency-band modulator. Experiments on ImageNet and MS-COCO demonstrate that UAE effectively unifies semantic abstraction and pixel-level fidelity, achieving state-of-the-art performance.