Browsing by Author "William, Albert"

Now showing 1 - 7 of 7
  • Thumbnail Image
    Item
    CNN-based network has Network Anisotropy -work harder to learn rotated feature than non-rotated feature
    (IEEE, 2022-10) Dale, Ashley S.; Qui, Mei; Christopher, Lauren; Krogg, Wen; William, Albert; Electrical and Computer Engineering, School of Engineering and Technology
    Successful object identification and classification in a generic Convolutional Neural Network (CNN) depends on object orientation. We expect CNN-based architectures to work harder to learn a rotated version of a feature than when learning the same feature in its default orientation. We name this phenomenon “Network Anisotropy”. A data set of 6000 RGB and grayscale images was created with rotated orientations of a feature predetermined and evenly distributed across four classes: 0°, 30°, 60°, 90°. Four ResNet (18, 34, 50, 101) classifier architectures were trained and the confidence scores were used to represent prediction accuracy. The results show that in all networks, training performance lags several epochs for the 30° and 60° rotation predictions compared to the 0° and 90° rotations, indicating a quantifiable network anisotropy. Because 0° and 90° both lie along a single rectilinear axis that coincides with the convolutional kernel of the CNN, we expect the classifier to do better on these two classes than on 30° and 60° classes. This work confirms that CNN architectures may have weaker performance based on feature orientation alone, independent of the feature distribution within the data set or the correlation of features within an image.
  • Thumbnail Image
    Item
    COMPARISON OF 3D VOLUME REGISTRATION TECHNIQUES APPLIED TO NEUROSURGERY
    (Office of the Vice Chancellor for Research, 2012-04-13) Verma, Romil; Cottingham, Chris; Nguyen, Thanh; Kale, Ashutosh; Catania, Robin; Wright, Jacob; Christopher, Lauren; Tuceryan, Mihan; William, Albert
    Introduction: Image guided surgery requires that the pre-operative da-ta used for planning the surgery should be aligned with the patient during surgery. For this surgical application a fast, effective volume registration al-gorithm is needed. In addition, such an algorithm can also be used to devel-op surgical training presentations. This research tests existing methods of image and volume registration with synthetic 3D models and with 3D skull data. The aim of this research is to find the most promising algorithms in ac-curacy and execution time that best fit the neurosurgery application. Methods: Medical image volumes acquired from MRI or CT medical im-aging scans provided by the Indiana University School of Medicine were used as Test image cases. Additional synthetic data with ground truth was devel-oped by the Informatics students. Each test image was processed through image registration algorithms found in four common medical imaging tools: MATLAB, 3D Slicer, VolView, and VTK/ITK. The resulting registration is com-pared against the ground truth evaluated with mean squared error metrics. Algorithm execution time is measured on standard personal computer (PC) hardware. Results: Data from this extensive set of tests reveal that the current state of the art algorithms all have strengths and weaknesses. These will be categorized and presented both in a poster form and in a 3D video presenta-tion produced by Informatics students in an auto stereoscopic 3D video. Conclusions: Preliminary results show that execution of image registra-tion in real-time is a challenging task for real time neurosurgery applica-tions. Final results will be available at paper presentation. Future research will focus on optimizing registration and also implementing deformable regis-tration in real-time.
  • Thumbnail Image
    Item
    Engineering and Informatics Student Multidisciplinary Learning using 3D Visualization and 3D Display of Radio Frequency (RF) Concepts
    (IEEE, 2018-10) Christopher, Lauren; William, Albert; Rao, Anusha S.; Dale, Ashley; Chase, Anthony; Joshi, Mihir Piyush; Krogg, Wendy; Abernathy, Bree; Electrical and Computer Engineering, School of Engineering and Technology
    This full paper addresses the Innovative Practice Category. We discuss our multidisciplinary approach to create a truly 3D representation and 3D display of RF signals in space through the development of two different training tools to enhance student understanding of Radio Communications. Both tools show the data on 3D autostereoscopic displays rather than rendered back to 2D displays. The first new tool is a series of 3D stereoscopic animations created by a multidisciplinary team of students from the Media Arts and Sciences (School of Informatics) and Electrical Engineering (School of Engineering) programs for use with an autostereoscopic display, where each animation focuses on a single topic within RF communication learning, using real-world examples. The second innovative tool models the Navy use-case of Electronic Warfare (EW) using examples with 3D antenna radiation patterns of signal propagation using U.S. Navy's SIMDIS interactive 3D visualization environment. The developed scenarios are displayed on an autostereoscopic display, allowing students to manipulate RF signals in a 3D environment. Learning gains were assessed via a 2x2 crossover experimental design an engineering student group. Compared to the control group, students showed gains in understanding of the 3D shape of dipole antennas and understanding of the multiple RF antennas in a cell phone, and the connections between mobile phone antennas and cell towers. The results from these interventions collectively indicate that a truly 3D representation in space can be used to enhance students' understanding of antennas and RF signals.
  • Thumbnail Image
    Item
    Greek Heritage: The Exploration of an Ancient Sanctuary
    (Office of the Vice Chancellor for Research, 2016-04-08) William, Albert; Lewis, Thomas; Townsend, Andrew
    ABSTRACT In the Summer of 2015, a team of faculty and students from IUPUI undertook the ambitious mission of traveling to the uninhabited Greek island of Despotiko to make the first cinematic documentary about an influential archeological discovery. This trip resulted in a thirteen-minute film about a wealthy and rebellious Greek island that challenged Athens for dominance in the ancient Greek world. This study abroad program provided students with a service learning experience that produced a significant documentation of Greek cultural heritage. On their base Island of Paros, the team collaborated with members of the community to create a modern production facility. Led by two faculty members from the IUPUI School of Informatics and Computing, a team of eleven students conducted research, interviewed local scholars, and created 3D recreations of an ancient Greek complex. In addition to their academic achievements, interaction with the broader community provided a critical component to their experience. The video, titled “Despotiko: An Unknown Treasure”, was well received locally and, in addition to being shown in the U.S., the film was debuted internationally in a Greek film festival. This combined effort laid the foundation of a new partnership with the residents of Paros and demonstrated how public scholarship can connect global communities towards mutual understanding and cooperation.
  • Thumbnail Image
    Item
    INTRINSIC DISORDER, SCAFFOLDS, AND STOCHASTIC MACHINES
    (Office of the Vice Chancellor for Research, 2010-04-09) Dunker, A. Keith; Xue, Bin; William, Albert; Uversky, Vladimir N.
    Scaffold proteins bind additional proteins that then carry out multi-step pathways. How do such machines work? Here a new hypothesis is proposed for the complex consisting of axin, two kinases – GSK3β and CK1α, and β-catenin. The pathway involves four discrete phosphorylations of β-catenin by the two kinases. Like many other scaffold proteins, axin is mostly unstructured [1, 2]. With a length of about 800 residues, axin forms two small domains of less than 100 residues each, and uses only a small number of residues, about 20 per interaction, to bind to GSK3β and β-catenin [1], and presumably also to bind to CK1α. Thus, even with the two domains and 3 partners, axin remains mostly unfolded. The hypothesis is that the unstructured axin molecule holds the three globular proteins in very high local concentrations, like three globules on a rope, and that, by random motions, first CK1α and then GSK3β phosphorylate the disordered tail of β-catenin successively four times. The “conformational changes” of axin that lead to acceleration of phosphorylation are neither specific nor coordinated, but rather are entirely stochastic, with stereochemical fit between the enzymes and their targets leading to the correct ordering of the four phosphorylation steps. In this hypothesis, the scaffold protein acts simply as a flexible tether that leads to acceleration of the multiple steps in the pathway by raising the local concentrations of the key components and by allowing the various components the freedom to collide in various orientations until productive collisions result. Thus, the steps of the pathway are carried out by a stochastic machine. This may be a general mechanism for scaffold-based molecular machines.
  • Thumbnail Image
    Item
    Technology as a Catalyst for Change: Engaging Audiences in the 21st Century
    (2015-08-16) Hook, Sara Anne; William, Albert
    This session showcases how 3D printing technology can be used to construct flutes that retain their traditional characteristics, but with 21st-century appeal that engages today’s audiences. It features a demonstration of the 3D-printed flutes and the system that created them.
  • Thumbnail Image
    Item
    Virtual Reality for Preservation: Production of Virtual Reality Heritage Spaces in the Classrooom
    (Council on Library and Information Resources, 2019) Wood, Zebulun M.; William, Albert; Copeland, Andrea; Human-Centered Computing, School of Informatics and Computing
    The Bethel AME Church was the oldest African American church in Indianapolis. In November 2016, the congregation moved out of downtown, and the building that had housed the congregation since 1869 was sold. It is now being redeveloped into a hotel. Through the Virtual Bethel project, faculty and students in the Media Arts and Science (MAS) program at Indiana University–Purdue University Indianapolis (IUPUI) created a 3D virtual space of the physical sanctuary to preserve the cultural heritage of Bethel. During its creation, Virtual Bethel served as a curricular and co-curricular experience for the undergraduate students in the 3D graphics and animation specialization within class N441 3D Team Production, which was co-taught by Albert William and Zebulun Wood. Virtual Bethel, finished in 2018, was the first historical and cultural preservation project that used VR within our class, program, school, and Indiana University (IU) campus. Users can interact with various types of primary sources (e.g., photographs, video, audio, text) to learn about the underrepresented history of African Americans associated with the church. Virtual Bethel was created in a series of classes within the MAS Program in the School of Informatics and Computing (SoIC), IUPUI. Methods of teaching a team of students to preserve historic spaces using VR are discussed, as are our philosophies toward productions when working with varying stakeholders’ priorities related to data preservation, asset preservation, and cultural preservation.