Repository logo

Browsing by Author "Azimi-Sadjadi, Mahmood, committee member"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Compartmentalization of membrane proteins by the actin cytoskeleton
    (Colorado State University. Libraries, 2013) Higgins, Jenny, author; Krapf, Diego, advisor; Tamkun, Michael, committee member; Bamburg, James, committee member; Azimi-Sadjadi, Mahmood, committee member
    Acting as the point of contact for the outside world, the plasma membrane is crucial for cellular signaling events. Proper organization of membrane components is necessary to accomplish this task. Although a number of experiments have demonstrated the compartmentalization of lipids and proteins on the plasma membrane, direct observation of the mechanisms by which the organization occurs has been challenging, in part due to the imaging restrictions of a diffraction-limited system and the dynamic nature of the membrane compartmentalization. Using photoactivated localization microscopy (PALM), a superresolution technique, we have captured the dynamics of compartments formed by the cortical actin cytoskeleton. Live human embryonic kidney (HEK293) cells were imaged with a temporal resolution of 2 s and a spatial resolution of 40 nm. The actin cytoskeleton forms compartments with a mean area of 2.3±0.3 μm2 that are partially outlined by actin bundles. When the PALM images of actin were combined with single particle tracking of membrane proteins, we directly observed the cytoskeleton acting as a barrier to the diffusion of Kv2.1 and Kv1.4, two voltage-gated potassium channels. In addition, we used a novel compartment detection and tracking algorithm to show that Kv2.1 and Kv1.4 channels avoid actin when changing compartments. This work represents the first direct observations of individual membrane protein interactions with barriers formed by the actin cytoskeleton.
  • Thumbnail Image
    ItemOpen Access
    Machine learning for computer aided programming: from stochastic program repair to verifiable program equivalence
    (Colorado State University. Libraries, 2022) Kommrusch, Steve, author; Pouchet, Louis-Noël, advisor; Anderson, Charles, advisor; Beveridge, Ross, committee member; Azimi-Sadjadi, Mahmood, committee member
    Computer programming has benefited from a virtuous cycle of innovation as improvements in computer hardware and software make higher levels of program abstraction and complexity possible. Recent advances in the field of machine learning, including neural network models for translating and answering questions about human language, can also be applied to computer programming itself. This thesis aims to make progress on the problem of using machine learning to improve the quality and robustness of computer programs by contributing new techniques for representation of programming problems, applying neural network models to code, and training procedures to create systems useful for computer aided programming. We first present background and preliminary studies of machine learning concepts. We then present a system that directly produces source code for automatic program repair which advances the state of the art by using a learned copy mechanism during generation. We extend a similar system to tune its learning for security vulnerability repair. We then develop a system for program equivalence which generates deterministically checkable output for equivalent programs. For this work we detail our contribution to the popular OpenNMT-py GitHub project used broadly for neural machine translation. Finally, we show how the deterministically checkable output can provide self-supervised sample selection which improves the performance and generalizability of the system. We develop breadth metrics to demonstrate that the range of problems addressed is representative of the problem space, while demonstrating that our deep neural networks generate proposed solutions which can be verified in linear time. Ultimately, our work provides promising results in multiple areas of computer aided programming which allow human developers to produce quality software more effectively.
  • Thumbnail Image
    ItemOpen Access
    Path planning for autonomous aerial vehicles using Monte Carlo tree search
    (Colorado State University. Libraries, 2024) Vasutapituks, Apichart, author; Chong, Edwin K. P., advisor; Azimi-Sadjadi, Mahmood, committee member; Pinaud, Olivier, committee member; Pezeshki, Ali, committee member
    Unmanned aerial vehicles (UAVs), or drones, are widely used in civilian and defense applications, such as search and rescue operations, monitoring and surveillance, and aerial photography. This dissertation focuses on autonomous UAVs for tracking mobile ground targets. Our approach builds on optimization-based artificial intelligence for path planning by calculating approximately optimal trajectories. This approach poses a number of challenges, including the need to search over large solution spaces in real-time. To address these challenges, we adopt a technique involving a rapidly-exploring random tree (RRT) and Monte Carlo tree search (MCTS). The RRT technique increases in computational cost as we increase the number of mobile targets and the complexity of the dynamics. Our MCTS approach executes a tree search based on random sampling to generate trajectories in real time. We develop a variant of MCTS for online path-planning to track ground targets together with an associated algorithm called P-UAV. Our algorithm is based on the framework of partially observable Monte Carlo planning, originally developed in the context of MCTS for Markov decision processes. Our real-time approach exploits a parallel-computing strategy with a heuristic random-sampling process. In our framework, We explicitly incorporate threat evasion, obstacle collision avoidance, and resilience to wind. The approach embodies an exploration-exploitation tradeoff in seeking a near-optimal solution in spite of the huge search space. We provide simulation results to demonstrate the effectiveness of our path-planning method.