Digital Commons Network^™

Wakening Past Concepts Without Past Data: Class-Incremental Learning From Online Placebos, Yaoyao Liu, Yingying Li, Bernt Schiele, Qianru Sun

Research Collection School Of Computing and Information Systems

Not forgetting old class knowledge is a key challenge for class-incremental learning (CIL) when the model continuously adapts to new classes. A common technique to address this is knowledge distillation (KD), which penalizes prediction inconsistencies between old and new models. Such prediction is made with almost new class data, as old class data is extremely scarce due to the strict memory limitation in CIL. In this paper, we take a deep dive into KD losses and find that "using new class data for KD"not only hinders the model adaption (for learning new classes) but also results in low efficiency for …

Demonstrating Canvas-Based Processing Of Multiple Camera Streams At The Edge, Ila Gokarn, Hemanth Sabbella, Yigong Hu, Tarek Abdelzaher, Archan Misra

Research Collection School Of Computing and Information Systems

We demonstrate criticality-aware canvas-based processing of multiple concurrent camera streams at the resource constrained edge to show substantial improvement in the accuracy-throughput trade-off. The proposed system focuses the available computation resources on select Regions of Interest (RoI) across all the camera streams by (i) extracting RoI from the input camera stream (ii) 2D bin packing the RoI on a canvas frame and (iii) batching and inferring upon these constructed composite canvas frames with a YOLOv5 object detection model. Our experiments show that such canvas-based processing can (i) sustain real-time processing throughput of 23 FPS per camera across 6 concurrent input …

Privobfnet: A Weakly Supervised Semantic Segmentation Model For Data Protection, Chiat Pin Tay, Vigneshwaran Subbaraju, Thivya Kandappu

Research Collection School Of Computing and Information Systems

The use of social media has made it easy to communicate and share information over the internet. However, it also brings issues such as data privacy leakage, which can be exploited by recipients with malicious intentions to harm the sender. In this paper, we propose a deep neural network that analyzes user’s image for privacy sensitive content and automatically locates sensitive regions for obfuscation. Our approach relies solely on image level annotations and learns to (a) predict an overall privacy score, (b) detect sensitive attributes and (c) demarcate the sensitive regions for obfuscation, in a given input image. We validated …

Societal Impacts Of Artificial Intelligence: Ethics, Legal, And Governance Issues, Yuzhou Qian, Keng Siau, Fiona Fui-Hoon Nah

Research Collection School Of Computing and Information Systems

Artificial intelligence (AI) is quickly changing the way we work and the way we live. The emergence of ChatGPT has thrust AI, especially Generative AI, into the spotlight. The societal impact of AI is on most people's minds. This article presents several research projects on how AI impacts work and society. Three research works are discussed in this article. The first study develops a theoretical framework structuring the legal and ethical objectives that are needed and the means to achieve them. The second study concentrates on bias and discrimination issues embedded in AI applications. It focuses on enhancing the collaboration …

A Secure And Robust Knowledge Transfer Framework Via Stratified-Causality Distribution Adjustment In Intelligent Collaborative Services, Ju Jia, Siqi Ma, Lina Wang, Yang Liu, Robert H. Deng

Research Collection School Of Computing and Information Systems

The rapid development of device-edge-cloud collaborative computing techniques has actively contributed to the popularization and application of intelligent service models. The intensity of knowledge transfer plays a vital role in enhancing the performance of intelligent services. However, the existing knowledge transfer methods are mainly implemented through data fine-tuning and model distillation, which may cause the leakage of data privacy or model copyright in intelligent collaborative systems. To address this issue, we propose a secure and robust knowledge transfer framework through stratified-causality distribution adjustment (SCDA) for device-edge-cloud collaborative services. Specifically, a simple yet effective density-based estimation is first employed to obtain …

Big Code Search: A Bibliography, Kisub Kim, Sankalp Ghatpande, Dongsun Kim, Xin Zhou, Kui Liu, Tegawende F. Bissyande, Jacques Klein, Traon Yves Le

Research Collection School Of Computing and Information Systems

Code search is an essential task in software development. Developers often search the internet and other code databases for necessary source code snippets to ease the development efforts. Code search techniques also help learn programming as novice programmers or students can quickly retrieve (hopefully good) examples already used in actual software projects. Given the recurrence of the code search activity in software development, there is an increasing interest in the research community. To improve the code search experience, the research community suggests many code search tools and techniques. These tools and techniques leverage several different ideas and claim a better …

Glance To Count: Learning To Rank With Anchors For Weakly-Supervised Crowd Counting, Zheng Xiong, Liangyu Chai, Wenxi Liu, Yongtuo Liu, Sucheng Ren, Shengfeng He

Research Collection School Of Computing and Information Systems

Crowd image is arguably one of the most laborious data to annotate. In this paper, we devote to reduce the massive demand of densely labeled crowd data, and propose a novel weakly-supervised setting, in which we leverage the binary ranking of two images with highcontrast crowd counts as training guidance. To enable training under this new setting, we convert the crowd count regression problem to a ranking potential prediction problem. In particular, we tailor a Siamese Ranking Network that predicts the potential scores of two images indicating the ordering of the counts. Hence, the ultimate goal is to assign appropriate …

Cooperative Trucks And Drones For Rural Last-Mile Delivery With Steep Roads, Jiuhong Xiao, Ying Li, Zhiguang Cao, Jianhua Xiao

Research Collection School Of Computing and Information Systems

The cooperative delivery of trucks and drones promises considerable advantages in delivery efficiency and environmental friendliness over pure fossil fuel fleets. As the prosperity of rural B2C e-commerce grows, this study intends to explore the prospect of this cooperation mode for rural last-mile delivery by developing a green vehicle routing problem with drones that considers the presence of steep roads (GVRPD-SR). Realistic energy consumption calculations for trucks and drones that both consider the impacts of general factors and steep roads are incorporated into the GVRPD-SR model, and the objective is to minimize the total energy consumption. To solve the proposed …

Clearspeech: Improving Voice Quality Of Earbuds Using Both In-Ear And Out-Ear Microphones, Dong Ma, Ting Dang, Ming Ding, Rajesh Krishna Balan

Research Collection School Of Computing and Information Systems

Wireless earbuds have been gaining increasing popularity and using them to make phone calls or issue voice commands requires the earbud microphones to pick up human speech. When the speaker is in a noisy environment, speech quality degrades significantly and requires speech enhancement (SE). In this paper, we present ClearSpeech, a novel deep-learningbased SE system designed for wireless earbuds. Specifically, by jointly using the earbud’s in-ear and out-ear microphones, we devised a suite of techniques to effectively fuse the two signals and enhance the magnitude and phase of the speech spectrogram. We built an earbud prototype to evaluate ClearSpeech under …

Enabling Ai And Robotic Coaches For Physical Rehabilitation Therapy: Iterative Design And Evaluation With Therapists And Post-Stroke Survivors, Min Hun Lee, Daniel Siewiorek, Asim Smailagic, Alexandre Bernardino, Sergi Bermúdez I Badia

Research Collection School Of Computing and Information Systems

Artificial intelligence (AI) and robotic coaches promise the improved engagement of patients on rehabilitation exercises through social interaction. While previous work explored the potential of automatically monitoring exercises for AI and robotic coaches, the deployment of these systems remains a challenge. Previous work described the lack of involving stakeholders to design such functionalities as one of the major causes. In this paper, we present our efforts on eliciting the detailed design specifications on how AI and robotic coaches could interact with and guide patient’s exercises in an effective and acceptable way with four therapists and five post-stroke survivors. Through iterative …

The Science Of Gravitational-Wave Sources And Beyond Compact Binary Coalescences, Yanyan Zheng

Doctoral Dissertations

"This work focuses on the field of gravitational-wave astronomy by extending the scope of detectable sources beyond compact binary coalescences, All the gravitational-wave detections so far come from compact binary coalescences. Focusing on core-collapse supernovae as promising sources for short gravitational-wave transients, this work reports optically targeted searches for gravitational-wave emitted by core-collapse supernovae during the third observing run of the LIGO and Virgo detectors. It also predicts the search sensitivity for the ongoing fourth and forthcoming fifth observing runs. Moreover, the work introduces a novel computational framework for testing the spatial distribution of binary black hole sources, allowing for …

Material Models For Simulation Of Steel Hot Rolling Process Using Experimentally Determined Properties At Elevated Temperatures And Strain Rates, Siva Sai Krishna Dasari

Doctoral Dissertations

"Hot rolling is one of the most important and complex process that includes large material deformation, different contact conditions, complex geometries and different temperature gradients along with micro structural changes such as viscoplasticity, dynamic recrystallization, and static softening. In the present work, the viscoplasticity, damage evolution, static softening and void closure of alloyed metal during hot rolling process was numerically investigated. Gleeble^TM hot compression and double hit tests were performed to provide experimental stress-strain curves at recrystallization temperatures and at different strain rates. Johnson-Cook viscoplastic model was calibrated using experimental stress-strain curves. Static softening model was developed at varying …

Deep Learning Techniques For Image Segmentation In Dermoscopic Skin Cancer Images, Norsang Lama

Doctoral Dissertations

"Melanoma is recognized as the most lethal type of skin cancer, responsible for a significant proportion of skin cancer-related deaths. However, early detection of melanoma is essential for successful treatment outcomes. Computer-aided skin cancer diagnosis tools can save lives by enabling earlier detection of skin cancer. Image segmentation is a crucial step in computer-aided diagnosis as it allows the detection of critical features or regions in an image. Thus, an accurate image segmentation method is necessary to create a more precise computer-aided diagnostic tool for skin cancer diagnosis. This dissertation includes investigating and developing deep learning techniques to improve image …

Improvements In Biomedical Image Analysis With Computational Intelligence And Data Fusion Techniques, Akanksha Maurya

Doctoral Dissertations

"An estimated 2 million new cases of basal cell carcinoma (BCC) are diagnosed each year in the United States, making it one of the most common skin cancers. Earlier detection of these cancers enables less invasive biopsies. Clinical detection consists of a preliminary visual observation of these skin lesions by an experienced dermatologist making it a specialized task highly dependent on their time, availability, and resources. Hence, there is a need for automating this process that can assist healthcare staff. In recent years, deep learning (DL) has been used extensively and successfully to diagnose different cancers in dermoscopic images. Telangiectasia …

Catalytic Control Of The Nanomorphology And Mechanical Properties Of Aliphatic Shape-Memory Aerogels, A B M Shaheen Ud Doulah

Doctoral Dissertations

"Shape-memory poly(isocyanurate-urethane) (PIR-PUR) aerogels are nanoporous solids that can return to their original shape after being compressed, through a heating actuation step. This thesis compares the effectiveness of various metal ions as catalysts in the formation of PIR-PUR aerogels, and explores the correlation between catalytic activity, nanomorphology, and mechanical properties of the resulting aerogels. The gelation rate was found to increase from Fe to Cu and then decline from Cu to Ga in the periodic table. CuCl₂ was found to be the fastest catalyst, and FeCl₃ the slowest. The morphology of the aerogels changed from bicontinuous to spheroidal …

Modeling And Control For Precision Robotic Machining, Patrick Bazzoli

Doctoral Dissertations

"Robots are used in a wide variety of manufacturing applications, but machining applications in which robots can excel are limited by their lower accuracy and stiffness relative to traditional CNC machines. This work is composed of two parts: one to evaluate a robot’s accuracy and one to compensate for the vibrations of the robot due to its lower stiffness.

In order to evaluate whether a robot has the necessary accuracy to perform a given machining task, Paper 1 discusses a novel Model Invalidation method. This methodology provides a statistical framework as well as a measurement strategy for determining if a …

Dynamics And Inverse Problems For Nonlinear Schrödinger Equations, Christopher Hogan

Doctoral Dissertations

"The cubic nonlinear Schrödinger equation (NLS) is a model of interest in the study of physical problems including nonlinear optics and Bose-Einstein condensates. Of particular interest is the study of cubic NLS with inhomogeneities such as localizations of the nonlinearity or terms introducing potential barriers. We first address some preliminaries and techniques useful in the study of the cubic NLS and its variations. We then consider the cubic NLS with a localized nonlinearity in dimensions d ≥ 2. We show that solutions with data given by small-amplitude wave packets accrue a nonlinear phase that determines the X-ray transform of the …

In-Depth Characterization Of The Selective Oxidation Products Of Hafnium Carbide, Jonathan Allen Scott

Doctoral Dissertations

"Materials capable of withstanding oxidative environments at temperatures >2000°C are necessary for hypersonic applications. Hafnium carbide (HfC) shows a combination of properties that lends itself toward these applications; however, its oxidation behavior is poorly understood. At 1300°C, HfC was found to oxidize via the following mechanism: 1) Initial oxidation of both C and Hf present in the system, 2) Formation of an interconnected pore network, whose size is related to the volume of gases being generated during the oxidation process, 3) Growth of the oxide scale to become sufficiently protective, 4) Formation of an initially amorphous “HfO₂C” at …

Deep Reinforcement Learning Based Strategies For Inverter Dominated Microgrids, Oroghene Oboreh-Snapps

Doctoral Dissertations

The integration of inverter-based distributed generators (IBDGs) in modern power systems has ushered in a new era of renewable energy utilization, prompting the rise of Microgrid (MG) architectures. Nevertheless, as IBDG penetration increases, a host of challenges surface. Foremost among these challenges are issues related to frequency stability stemming from the inherent lack of inertia in IBDGs, and inaccurate sharing of reactive and active power, particularly evident in parallel IBDG networks with mismatched feeder impedance. This research presents novel solutions to these challenges by adopting deep reinforcement learning (DRL), specifically, the twin delayed deep deterministic policy gradient (TD3) algorithm, to …

Naming, Expressive Interests, And The Law: The Implications Of Governmental Form Design, Laura A. Heymann

Faculty Publications

The design of governmental forms can result in restrictions on the ability of individuals to submit truthful personal naming choices, such as when a form field has character limitations or does not accept diacritics. In some instances, form design operationalizes government policy. But in other instances, form design can itself create law by limiting the ability to submit accurate names on the front end. Awareness of the legal implications of form design is critically important. A more inclusive and accurate approach toward the legal documentation of naming choices means that forms should adapt to people, not the other way around.

Electrostatic Sorting Of Lunar Regolith Simulants For Sustainable Resource Utilization: Modeling And Characterization Of Particle Size Distribution, Abdullah Al Moinee, Peter Bachle, Kyle Newport, William Schonberg, David Bayless, Jeffrey Smith, Daoru Han, Fateme Rezaei

Civil, Architectural and Environmental Engineering Faculty Research & Creative Works

In pursuit of sustainable resource utilization on the Moon, this paper delves into modeling and characterization of particle size distribution (PSD) of lunar regolith simulants in an electrostatic system. a prototype electrostatic sieve was built and tested with four sample simulants mirroring properties of lunar mare and highland regolith. an alternating four-phase (90 degrees, 180 degrees, 270 degrees, 360 degrees) traveling square-wave was utilized for particle-directed transport to model the diverse trajectories of the particles. Numerically, we scrutinized how the distribution functions of the particles are manifested as the electrostatic field propagates, with a focus on three distinct particle ranges …

Token-Based Reviewer Economies: Proposed Institutions For Managing The Reviewer Shortage Problem, Cecil Eng Huang Chua

Business and Information Technology Faculty Research & Creative Works

This opinion paper presents two proposed token-based systems to fix the information system academy's review system. At present, the review system consumes more human resources than the information systems academy has by an order of magnitude. The cost of this overflow is borne entirely by information systems researchers. I argue this is because the review system is based on a gift economy that cannot handle the currently sized market. Both proposals involve the creation of tokens and a central governing body. The first proposal involves a market built around a review bank (central governing body) that issues review tokens that …

A Comparative Study Of Cationic Copper(I) Reagents Supported By Bipodal Tetramethylguanidinyl-Containing Ligands As Nitrene-Transfer Catalysts, Suraj Kumar Sahoo, Brent Harfmann, Himanshu Bhatia, Harish Singh, Srikanth Balijapelly, Amitava Choudhury, Pericles Stavropoulos

Chemistry Faculty Research & Creative Works

The Bipodal Compounds [(TMG2biphenN-R)CuI-NCMe](PF6) (R = Me, Ar (4-CF3Ph-)) And [(TMG2biphenN-Me)CuI-I] Have Been Synthesized With Ligands That Feature A Diarylmethyl- And Triaryl-Amine Framework And Superbasic Tetramethylguanidinyl Residues (TMG). The Cationic Cu(I) Sites Mediate Catalytic Nitrene-Transfer Reactions Between The Imidoiodinane PhI = NTs (Ts = Tosyl) And A Panel Of Styrenes In MeCN, To Afford Aziridines, Demonstrating Comparable Reactivity Profiles. The Copper Reagents Have Been Further Explored To Execute C-H Amination Reactions With A Variety Of Aliphatic And Aromatic Hydrocarbons And Two Distinct Nitrene Sources PhI = NTs And PhI = NTces (Tces = 2,2,2-Trichloroethylsulfamate) In Benzene/HFIP (10:2 V/v). Good Yields …

Perfect Polar Alignment Of Parallel Beloamphiphile Layers: Improved Structural Design Bias Realized In Ferroelectric Crystals Of The Novel “Methoxyphenyl Series Of Acetophenone Azines”, Harmeet Bhoday, Nathan Knotts, Rainer Glaser

Chemistry Faculty Research & Creative Works

An Improved Design Is Described For Ferroelectric Crystals And Implemented With The "Methoxyphenyl Series" Of Acetophenone Azines, (MeO−Ph, Y)-Azines With Y=F (1), Cl (2), Br (3), Or I (4). The Crystal Structures Of These Azines Exhibit Polar Stacking Of Parallel Beloamphiphile Monolayers (PBAMs). Azines 1, 3, And 4 Form True Racemates Whereas Chloroazine 2 Crystallizes As A Kryptoracemate. Azines 1–4 Are Helical Because Of The N−N Bond Conformation. In True Racemates The Molecules Of Opposite Helicity (M And P) Are Enantiomers A(M) And A*(P) While In Kryptoracemates They Are Diastereomers A(M) And B*(P). The Stacking Mode Of PBAMs Is Influenced …

Solar Enhanced Oxygen Evolution Reaction With Transition Metal Telluride, Harish Singh, Taishi Higuchi-Roos, Fabrice Roncoroni, David Prendergast, Manashi Nath

Chemistry Faculty Research & Creative Works

The Photo-Enhanced Electrocatalytic Method of Oxygen Evolution Reaction (OER) Shows Promise for Enhancing the Effectiveness of Clear Energy Generation through Water Splitting by using Renewable and Sustainable Source of Energy. However, Despite Benefits of Photo electrocatalytic (PEC) Water Splitting, its Uses Are Constrained by its Low Efficiency as a Result of Charge Carrier Recombination, a Large overpotential, and Sluggish Reaction Kinetics. Here, We Illustrate that Nickel Telluride (NiTe) Synthesized by Hydrothermal Methods Can Function as an Extremely Effective Photo-Coupled Electrochemical Oxygen Evolution Reaction (POER) Catalyst. in This Study, NiTe Was Synthesized by Hydrothermal Method at 145°C within Just an Hour …

Synthesizing And Using Iron Oxide Nanoparticles As Nanocomposite In Cotton Fabrics Nanofinishing, Hoda Sharouf, Ziad Saffour

Baghdad Science Journal

Metal oxide nanoparticles, including iron oxide, are highly considered as one of the most important species of nanomaterials in a varied range of applications due to their optical, magnetic, and electrical properties. Iron oxides are common compounds, extensive in nature, and easily synthesized in the laboratory. In this paper, iron oxide nanoparticles were prepared by co-precipitation of (Fe+2) and (Fe+3) ions, using iron (II and III) sulfate as precursor material and NH4OH solution as solvent at 90°C. After the synthesis of iron oxide particles, it was characterized using X-ray diffraction (XRD), infrared spectroscopy (FTIR), and scanning electron microscopy (SEM). These …

Classification Of Arabic Alphabets Using A Combination Of A Convolutional Neural Network And The Morphological Gradient Method, Mouhssine El Atillah, Khalid El Fazazy, Jamal Riffi

Baghdad Science Journal

The field of Optical Character Recognition (OCR) is the process of converting an image of text into a machine-readable text format. The classification of Arabic manuscripts in general is part of this field. In recent years, the processing of Arabian image databases by deep learning architectures has experienced a remarkable development. However, this remains insufficient to satisfy the enormous wealth of Arabic manuscripts. In this research, a deep learning architecture is used to address the issue of classifying Arabic letters written by hand. The method based on a convolutional neural network (CNN) architecture as a self-extractor and classifier. Considering the …

Semi-Analytical Assessment Of Magneto-Hydrodynamic Nano-Fluid Flow Jeffrey- Hamel Problem, Saja I Abdulridah, Abeer M Jasim

Baghdad Science Journal

In this paper, analyzing the non-dimensional Magnesium-hydrodynamics problem Using nanoparticles in Jeffrey-Hamel flow (JHF) has been studied. The fundamental equations for this issue are reduced to a three-order ordinary differential equation. The current project investigated the effect of the angles between the plates, Reynolds number, nanoparticles volume fraction parameter, and magnetic number on the velocity distribution by using analytical technique known as a perturbation iteration scheme (PIS). The effect of these parameters is similar in the converging and diverging channels except magnetic number that it is different in the divergent channel. Furthermore, the resulting solutions with good convergence and high …

Enhanced Photocatalytic Activity Of Cu2o/Zno/Go Nanocomposites On The Methylene Blue Degradation, Frikson Jony Purba, Zuriah Sitorus, Kerista Tarigan, Nurdin Siregar

Baghdad Science Journal

This study synthesized nanocomposite photocatalyst materials from a mixture of Cu2O nanoparticles, ZnO nanoparticles, and graphene oxide (GO) through coprecipitation and hydrothermal methods. This study aims to determine the optimum composition of Cu2O/ZnO/GO nanocomposites in degrading methylene blue. The nanocomposite was synthesized in two steps: 1 the synthesis of Cu2O and ZnO nanoparticles through the coprecipitation method and the preparation of GO through the modified Hummer method. 2 The preparation of Cu2O and ZnO nanoparticles mixtures with GO through the hydrothermal method to form Cu2O/ZnO/GO nanocomposites. The adsorption-photocatalysis process of methylene blue was done with UV light from a halogen …

Polar Stacking Of Dipole Parallel-Aligned Monolayers Of Unsymmetrical 1,4-Diphenyl-1,3-Butadienes Creates Nonlinear Optical Materials: Insights From Experiments Guide Structure Assignments, Harmeet Bhoday, Justin Nulsen, Steven P. Kelley, Rainer Glaser

Chemistry Faculty Research & Creative Works

A series of 1-(4′-halophenyl)-4-(4″-methoxyphenyl) buta-1,3-dienes, (H, MeO, Y)-1,4-diphenylbutadienes with halogens Y = F (1), Cl (2), Br (3), and I (4) are described. Crystal structure analysis establishes that 1-4 present a new class of highly dipole parallel-aligned polar organic molecular materials for nonlinear optics (NLO). Building on previous studies of polar crystals of unsymmetrical acetophenone azines, (R, X, Y)-azines, X-Ph-CR═N-N═CR-Ph-Y (R = Me; X = R'O, PhO, R'OPh; Y = F, Cl, Br, I), we suspected that the all-carbon analogs (R, R'O, Y)-1,4-diphenylbutadienes, R'O-Ph-CR═CH-CH═CR-Ph-Y, could crystallize with polar lattices and feature stronger NLO effects because of their improved conjugation and …