Congratulations to our two Top Scholars!
Over the past Ten years, Eleven Advanced Math/Science Research students have been named Science Talent Search (STS) semifinalists, a distinction earned by only 300 students in the country each year.
RENAMED THE REGENERON STS (FORMERLY INTEL) IN 2017, IT IS THE MOST PRESTIGIOUS SCIENCE COMPETITION FOR PRE-COLLEGE STUDENTS IN THE NATION. THE STS has been CALLED THE "JUNIOR NOBEL"
Propionibacterium acnes biofilm: The Effect of Peptide 1018 on Bacterial Production
Avalon Z. M. Lebenthal
Abstract: Propionibacterium acnes is a major component of the human skin microbiome and plays an important role in the etiology of acne. P. acnes produces a biofilm, aiding in its growth and protection. Researchers studying the destruction and growth of biofilm have recently linked peptide 1018 to the breakdown of many different bacterial biofilms (and its prevention of reformation of films in these colonies). This peptide works to disrupt biofilm growth by blocking the cellular production of (p)ppGpp in the organisms. In this study, the presence of biofilm was investigated in various amounts of peptide 1018; studied through digital pixelation density recordings of stained microtiter plates and electron microscopy. Peptide 1018 was effective in preventing P. acne biofilm. After peptide exposure, the P. acnes may have increased sensitivity to various anti-microbial products, including silver, copper, sulfur, blue light phototherapy, and manuka honey. The results from the peptide effectiveness tests could later be used with these substances for the development of a combination therapy that will kill off the bacteria, without harming human tissue.
Robust Adversarial Perturbation in Deep Proposal-based Models
Making driverless cars even safer....
Abstract: Adversarial noises are useful tools to probe the weakness of deep learning-based computer vision algorithms. This research describes a robust adversarial perturbation (R-AP) to attack deep proposal-based object detectors and instance segmentation algorithms. The proposed method focuses on attacking the common component in these algorithms, namely Region Proposal Network (RPN), to universally degrade their performances in a black-box fashion. To do so, a loss function, which combines a label loss and a novel shape loss, was optimized with respect to an image using a gradient based iterative algorithm. Evaluations are performed on the MS COCO 2014 dataset for the adversarial attacking of 6 state-of-the-art object detectors and 2 instance segmentation algorithms. Experimental results demonstrate the efficacy of the proposed method.