Computer Sciences Commons^™

Training Thinner And Deeper Neural Networks: Jumpstart Regularization, Carles Riera, Camilo Rey, Thiago Serra, Eloi Puertas, Oriol Pujol

Faculty Conference Papers and Presentations

Neural networks are more expressive when they have multiple layers. In turn, conventional training methods are only successful if the depth does not lead to numerical issues such as exploding or vanishing gradients, which occur less frequently when the layers are sufficiently wide. However, increasing width to attain greater depth entails the use of heavier computational resources and leads to overparameterized models. These subsequent issues have been partially addressed by model compression methods such as quantization and pruning, some of which relying on normalization-based regularization of the loss function to make the effect of most parameters negligible. In this work, …

Monofacial Vs Bifacial Solar Photovoltaic Systems In Snowy Environments, Koami Soulemane Hayibo, Aliaksei Petsiuk, Pierce Mayville, Laura Brown, Joshua M. Pearce

Electrical and Computer Engineering Publications

There has been a recent surge in interest in the more accurate snow loss estimates for solar photovoltaic (PV) systems as large-scale deployments move into northern latitudes. Preliminary results show bifacial modules may clear snow faster than monofacial PV. This study analyzes snow losses on these two types of systems using empirical hourly data including energy, solar irradiation and albedo, and open-source image processing methods from images of the arrays in a northern environment in the winter. Projection transformations based on reference anchor points and snowless ground truth images provide reliable masking and optical distortion correction with fixed surveillance cameras. …

Improving Pain Assessment Using Vital Signs And Pain Medication For Patients With Sickle Cell Disease: Retrospective Study, Swati Padhee, Gary K. Nave Jr, Tanvi Banerjee, Daniel M. Abrams, Nirmish Shah

Computer Science and Engineering Faculty Publications

Background: Sickle cell disease (SCD) is the most common inherited blood disorder affecting millions of people worldwide. Most patients with SCD experience repeated, unpredictable episodes of severe pain. These pain episodes are the leading cause of emergency department visits among patients with SCD and may last for several weeks. Arguably, the most challenging aspect of treating pain episodes in SCD is assessing and interpreting a patient's pain intensity level. Objective: This study aims to learn deep feature representations of subjective pain trajectories using objective physiological signals collected from electronic health records. Methods: This study used electronic health record data collected …

How To Detect (And Analyze) Independent Subsystems Of A Black-Box (Or Grey-Box) System, Saeid Tizpaz-Niari, Olga Kosheleva, Vladik Kreinovich

Departmental Technical Reports (CS)

Often, we deal with black-box or grey-box systems where we can observe the overall system's behavior, but we do not have access to the system's internal structure. In many such situations, the system actually consists of two (or more) independent components: a) how can we detect this based on observed system's behavior? b) what can we learn about the independent subsystems based on the observation of the system as a whole? In this paper, we provide (partial) answers to these questions.

Hawthorne Effect: An Explanation Based On Decision Theory, Sofia Holguin, Vladik Kreinovich, Nguyen Hoang Phuong

Departmental Technical Reports (CS)

It is known that people feel better (and even work better) if someone pays attention to them; this is known as the Hawthorne effect. At first glance, it sounds counter-intuitive: this attention does not bring you any material benefits, so why would you feel better? If you are sick and someone gives you medicine, this will make you feel better, but if someone just pays attention, why does that make you feel better? In this paper, we use the general ideas of decision theory to explain this seemingly counterintuitive phenomenon.

Why Should Exactly 1/4 Be Returned To The Original Owner: An Economic Explanation Of An Ancient Recommendation, Olga Kosheleva, Vladik Kreinovich

Departmental Technical Reports (CS)

What if someone bought a property in good faith, not realizing that this property was unjustly confiscated from the previous owner? In such situations, if the new owner decided to sell this property, Talmud recommended that a fair way is to return 1/4 of the selling price to the original owner. However, it does not provide any explanation of why exactly 1/4 -- and not any other portion -- is to be returned. In this paper, we provide an economic explanation for this recommendation, an explanation that fits well with other ancient recommendations about debts.

Why Convex Combination Is An Effective Crossover Operation In Continuous Optimization: A Theoretical Explanation, Kelly Cohen, Olga Kosheleva, Vladik Kreinovich

Departmental Technical Reports (CS)

When evolutionary computation techniques are used to solve continuous optimization problems, usually, convex combination is used as a crossover operation. Empirically, this crossover operation works well, but this success is, from the foundational viewpoint, a challenge: why this crossover operation works well is not clear. In this paper, we provide a theoretical explanation for this empirical success.

Estimating Skewness And Higher Central Moments Of An Interval-Valued Fuzzy Set, Juan Carlos Figueroa-Garcia, Martine Ceberio, Olga Kosheleva, Vladik Kreinovich

Departmental Technical Reports (CS)

A known relation between membership functions and probability density functions allows us to naturally extend statistical characteristics like central moments to the fuzzy case. In case of interval-valued fuzzy sets, we have several possible membership functions consistent with our knowledge. For different membership functions, in general, we have different values of the central moments. It is therefore desirable to compute, in the interval-valued fuzzy case, the range of possible values for each such moment. In this paper, we provide efficient algorithms for this computation.

Multiple Object Tracking For Marine Science, Nicholas A. Wachter

Computer Science and Software Engineering

As drone and computer vision technology has been improving, many fields of study have been quick to utilize it to improve the accuracy and ease of data collection. The combination of the two technologies is perfect for surveying large areas and identifying features of interest. Marine science utilizes these technologies for activities such as animal tracking and population counting. I am working with the Drones for Marine Science research group at Cal Poly who want to build a fleet of drones that will fly out over the ocean to identify and track various marine animals. My role will be to …

A Survey On Visual Analysis Of Massive Open Online Courses (Mooc) Data [大型开放式在线课程(Mooc)数据可视分析综述], Zhiguang Zhou, Saqing Wang, Yuhua Liu, Yong Wang, Yigang Wang

Research Collection School Of Computing and Information Systems

Massive open online courses(MOOC)are free online courses that provide educational resources to users.In this paper,related works are summarized from the following perspectives:data types of MOOC,MOOC data visualization,and visual analysis applications of MOOC data.First,MOOC data is categorized into four types,including text data,video data,topological data,spatio-temporal data,and their concepts and applications are introduced.Then,the common visualization methods of different data types are discussed.Furthermore,the major applications of visual analytics approaches in MOOC data analysis are summarized,including online course learning,online tests and exams,online forum interaction and so on.Finally,the challenges and future work on visual analytics of MOOC data are discussed and concluded.

Reinforcement Learning-Based Interactive Video Search, Zhixin Ma, Jiaxin Wu, Zhijian Hou, Chong-Wah Ngo

Research Collection School Of Computing and Information Systems

Despite the rapid progress in text-to-video search due to the advancement of cross-modal representation learning, the existing techniques still fall short in helping users to rapidly identify the search targets. Particularly, in the situation that a system suggests a long list of similar candidates, the user needs to painstakingly inspect every search result. The experience is frustrated with repeated watching of similar clips, and more frustratingly, the search targets may be overlooked due to mental tiredness. This paper explores reinforcement learning-based (RL) searching to relieve the user from the burden of brute force inspection. Specifically, the system maintains a graph …

Group Contextualization For Video Recognition, Yanbin Hao, Hao Zhang, Chong-Wah Ngo, Xiangnan He

Research Collection School Of Computing and Information Systems

Learning discriminative representation from the complex spatio-temporal dynamic space is essential for video recognition. On top of those stylized spatio-temporal computational units, further refining the learnt feature with axial contexts is demonstrated to be promising in achieving this goal. However, previous works generally focus on utilizing a single kind of contexts to calibrate entire feature channels and could hardly apply to deal with diverse video activities. The problem can be tackled by using pair-wise spatio-temporal attentions to recompute feature response with cross-axis contexts at the expense of heavy computations. In this paper, we propose an efficient feature refinement method that …

Multi-View Scheduling Of Onboard Live Video Analytics To Minimize Frame Processing Latency, Shengzhong Liu, Tianshi Wang, Hongpeng Guo, Xinzhe Fu, Philip David, Maggie Wigness, Archan Misra, Tarek Abdelzaher

Research Collection School Of Computing and Information Systems

This paper presents a real-time multi-view scheduling framework for DNN-based live video analytics at the edge to minimize frame processing latency. The work is motivated by applications where a higher frame rate is important, not to miss actions of interest. Examples include defense, border security, and intruder detection applications where sensors (in this paper, cameras) are deployed to monitor key roads, chokepoints, or passageways to identify events of interest (and intervene in real-time). Supporting a higher frame rate entails lowering frame processing latency. We assume that multiple cameras are deployed with partially overlapping views. Each camera has access to limited …

A Multi-Criteria Decision-Making (Mcdm) Approach For Data-Driven Distance Learning Recommendations, Aysha Meshaal Alshamsi

Theses

Distance learning has been adopted as an alternative learning strategy to the dominant face-to-face teaching methodology. It has been largely implemented by many governments worldwide due to the spread of the COVID-19 pandemic and the implication in enforcing lockdown and social distancing. In emergency situations distance learning is referred to as Emergency Remote Teaching (ERT). Due to this dynamic, sudden shift, and scaling demand in distance learning, many challenges have been accentuated. These include technological adoption, student commitments, parent involvement, and teacher extra burden management, changes in the organization methodology, in addition to government development of new guidelines and regulations …

Comparing Learned Representations Between Unpruned And Pruned Deep Convolutional Neural Networks, Parker Mitchell

Master's Theses

While deep neural networks have shown impressive performance in computer vision tasks, natural language processing, and other domains, the sizes and inference times of these models can often prevent them from being used on resource-constrained systems. Furthermore, as these networks grow larger in size and complexity, it can become even harder to understand the learned representations of the input data that these networks form through training. These issues of growing network size, increasing complexity and runtime, and ambiguity in the understanding of internal representations serve as guiding points for this work.

In this thesis, we create a neural network that …

Rasm: Compiling Racket To Webassembly, Grant Matejka

Master's Theses

WebAssembly is an instruction set designed for a stack based virtual machine, with an emphasis on speed, portability and security. As the use cases for WebAssembly grow, so does the desire to target WebAssembly in compilation. In this thesis we present Rasm, a Racket to WebAssembly compiler that compiles a select subset of the top forms of the Racket programming language to WebAssembly. We also present our early findings in our work towards adding a WebAssembly backend to the Chez Scheme compiler that is the backend of Racket. We address initial concerns and roadblocks in adopting a WebAssembly backend and …

Scheduling For Space Tracking And Heterogeneous Sensor Environments, Gabriel H. Greve

Theses and Dissertations

This dissertation draws on the fields of heuristic and meta-heuristic algorithm development, resource allocation problems, and scheduling to address key Air Force problems. The world runs on many schedules. People depend upon them and expect these schedules to be accurate. A process is needed where schedules can be dynamically adjusted to allow tasks to be completed efficiently. For example, the Space Surveillance Network relies on a schedule to track objects in space. The schedule must use sensor resources to track as many high-priority satellites as possible to obtain orbit paths and to warn of collision paths. Any collisions that occurred …

Legislative Language For Success, Sanjana Gundala

Master's Theses

Legislative committee meetings are an integral part of the lawmaking process for local and state bills. The testimony presented during these meetings is a large factor in the outcome of the proposed bill. This research uses Natural Language Processing and Machine Learning techniques to analyze testimonies from California Legislative committee meetings from 2015-2016 in order to identify what aspects of a testimony makes it successful. A testimony is considered successful if the alignment of the testimony matches the bill outcome (alignment is "For" and the bill passes or alignment is "Against" and the bill fails). The process of finding what …

Developing A Miniature Smart Boat For Marine Research, Michael Isaac Eirinberg

Computer Engineering

This project examines the development of a smart boat which could serve as a possible marine research apparatus. The smart boat consists of a miniature vessel containing a low-cost microcontroller to live stream a camera feed, GPS telemetry, and compass data through its own WiFi access point. The smart boat also has the potential for autonomous navigation. My project captivated the interest of several members of California Polytechnic State University, San Luis Obispo’s (Cal Poly SLO) Marine Science Department faculty, who proposed a variety of fascinating and valuable smart boat applications.

A Tale Of Two Regulatory Regimes: Creation And Analysis Of A Bilingual Privacy Policy Corpus, Siddhant Arora, Henry Hosseini, Christine Utz, Vinayshekhar Bannihatti Kumar, Jasmine Mangat, Peter Story, Rex Chen, Martin Degeling, Tom Norton, Thomas Hupperich

Computer Science

Over the past decade, researchers have started to explore the use of NLP to develop tools aimed at helping the public, vendors, and regulators analyze disclosures made in privacy policies. With the introduction of new privacy regulations, the language of privacy policies is also evolving, and disclosures made by the same organization are not always the same in different languages, especially when used to communicate with users who fall under different jurisdictions. This work explores the use of language technologies to capture and analyze these differences at scale. We introduce an annotation scheme designed to capture the nuances of two …

Ethical Implications For Children’S Use Of Search Tools In An Educational Setting, Monica Landoni, Theo Huibers, Emiliana Murgia, Maria Soledad Pera

Computer Science Faculty Publications and Presentations

In the classroom, search tools enable students to access online resources. While these tools have many benefits in theory, in practice there are also ethical issues to consider. In this article, we discuss a number of ethics-related problems teachers are faced with and they need to find solutions for. Based on our own research experience developing and deploying information discovery tools for the classroom (both in a traditional classroom setting and on the Internet due to the ongoing outbreak of COVID-19), we share insights about ethics and the role of the expert-in-the-loop, teachers, both as co-design partners and liaisons between …

Gauging The Acceptance Of Contact Tracing Technology: An Empirical Study Of Singapore Residents’ Concerns With Sharing Their Information And Willingness To Trust, Ee-Ing Ong, Wee Ling Loo

Research Collection Yong Pung How School Of Law

In response to the COVID-19 pandemic, governments began implementing various forms of contact tracing technology. Singapore’s implementation of its contact tracing technology, TraceTogether, however, was met with significant concern by its population, with regard to privacy and data security. This concern did not fit with the general perception that Singaporeans have a high level of trust in its government. We explore this disconnect, using responses to our survey (conducted pre-COVID-19) in which we asked participants about their level of concern with the government and business collecting certain categories of personal data. The results show that respondents had less concern with …

Consensus Embedding For Multiple Networks: Computation And Applications, Mengzhen Li, Mehmet Koyutürk

Faculty Scholarship

Machine learning applications on large-scale network-structured data commonly encode network information in the form of node embeddings. Network embedding algorithms map the nodes into a low-dimensional space such that the nodes that are “similar” with respect to network topology are also close to each other in the embedding space. Real-world networks often have multiple versions or can be “multiplex” with multiple types of edges with different semantics. For such networks, computation of Consensus Embeddings based on the node embeddings of individual versions can be useful for various reasons, including privacy, efficiency, and effectiveness of analyses. Here, we systematically investigate the …

Leveraging Context Patterns For Medical Entity Classification, Garrett Johnston

Computer Science Senior Theses

The ability of patients to understand health-related text is important for optimal health outcomes. A system that can automatically annotate medical entities could help patients better understand health-related text. Such a system would also accelerate manual data annotation for this low-resource domain as well as assist in down- stream medical NLP tasks such as finding textual similarity, identifying conflicting medical advice, and aspect-based sentiment analysis. In this work, we investigate a state-of-the-art entity set expansion model, BootstrapNet, for the task of medical entity classification on a new dataset of medical advice text. We also propose EP SBERT, a simple model …

Symplectically Integrated Symbolic Regression Of Hamiltonian Dynamical Systems, Daniel Dipietro

Computer Science Senior Theses

Here we present Symplectically Integrated Symbolic Regression (SISR), a novel technique for learning physical governing equations from data. SISR employs a deep symbolic regression approach, using a multi-layer LSTMRNN with mutation to probabilistically sample Hamiltonian symbolic expressions. Using symplectic neural networks, we develop a model-agnostic approach for extracting meaningful physical priors from the data that can be imposed on-the-fly into the RNN output, limiting its search space. Hamiltonians generated by the RNN are optimized and assessed using a fourth-order symplectic integration scheme; prediction performance is used to train the LSTM-RNN to generate increasingly better functions via a risk-seeking policy gradients …

Class Re-Activation Maps For Weakly-Supervised Semantic Segmentation, Zhaozheng Chen, Tan Wang, Xiongwei Wu, Xian-Sheng Hua, Hanwang Zhang, Qianru Sun

Research Collection School Of Computing and Information Systems

Extracting class activation maps (CAM) is arguably the most standard step of generating pseudo masks for weakly supervised semantic segmentation (WSSS). Yet, we find that the crux of the unsatisfactory pseudo masks is the binary cross-entropy loss (BCE) widely used in CAM. Specifically, due to the sum-over-class pooling nature of BCE, each pixel in CAM may be responsive to multiple classes co-occurring in the same receptive field. To this end, we introduce an embarrassingly simple yet surprisingly effective method: Reactivating the converged CAM with BCE by using softmax crossentropy loss (SCE), dubbed ReCAM. Given an image, we use CAM to …

Mlp-3d: A Mlp-Like 3d Architecture With Grouped Time Mixing, Zhaofan Qiu, Ting Yao, Chong-Wah Ngo, Tao Mei

Research Collection School Of Computing and Information Systems

Convolutional Neural Networks (CNNs) have been re-garded as the go-to models for visual recognition. More re-cently, convolution-free networks, based on multi-head self-attention (MSA) or multi-layer perceptrons (MLPs), become more and more popular. Nevertheless, it is not trivial when utilizing these newly-minted networks for video recognition due to the large variations and complexities in video data. In this paper, we present MLP-3D networks, a novel MLP-like 3D architecture for video recognition. Specifically, the architecture consists of MLP-3D blocks, where each block contains one MLP applied across tokens (i.e., token-mixing MLP) and one MLP applied independently to each token (i.e., channel MLP). …

Coded Distributed Function Computation, Pedro J. Soto

Dissertations, Theses, and Capstone Projects

A ubiquitous problem in computer science research is the optimization of computation on large data sets. Such computations are usually too large to be performed on one machine and therefore the task needs to be distributed amongst a network of machines. However, a common problem within distributed computing is the mitigation of delays caused by faulty machines. This can be performed by the use of coding theory to optimize the amount of redundancy needed to handle such faults. This problem differs from classical coding theory since it is concerned with the dynamic coded computation on data rather than just statically …

Determining Knowledge From Student Performance Prediction Using Machine Learning, Wala El Rashied Mohamed

Theses

Recent years have seen a rapid development in the field of educational data mining (EDM), enhancing the ability to trace student knowledge. Data from intelligent tutoring systems (ITS) have been analyzed and interpreted by multiple researchers seeking to measure students’ knowledge as it evolves. Human nature, as well as other factors, makes it difficult to determine whether or not students are knowledgeable. This thesis sets out to examine the level of students’ knowledge by predicting their current and future academic performance based on records of their historical interactions. By restructuring data and considering a student perspective, we can gain insight …

Single-Pass Inline Pipeline 3d Reconstruction Using Depth Camera Array, Zhexiong Shang, Zhigang Shen

Department of Construction Engineering and Management: Faculty Publications

A novel inline inspection (ILI) approach using depth cameras array (DCA) is introduced to create high-fidelity, dense 3D pipeline models. A new camera calibration method is introduced to register the color and the depth information of the cameras into a unified pipe model. By incorporating the calibration outcomes into a robust camera motion estimation approach, dense and complete 3D pipe surface reconstruction is achieved by using only the inline image data collected by a self-powered ILI rover in a single pass through a straight pipeline. The outcomes of the laboratory experiments demonstrate one-millimeter geometrical accuracy and 0.1-pixel photometric accuracy. …