41530 research outputs found
Sort by
AIOPS–Driven Adaptive Anomaly Detection in Evolving Cloud Environments Using Transfer Learning
Full text linkAs cloud-based microservice architectures have become the foundation of contempo- rary enterprise solutions, performance interference, wherein co-located services com- pete for shared resources, remains a significant challenge. This phenomenon, often referred to as the noisy neighbor problem, manifests when one workload unexpect- edly increases the CPU, memory, disk I/O, or network consumption, resulting in latency spikes or throughput degradation for other services. While existing isolation mechanisms (e.g., cgroups and QoS policies) provide some mitigation, they rarely prevent contention entirely, particularly in dynamic, rapidly evolving environments with frequent code deployments.
This thesis proposes an AIOps-driven adaptive anomaly detection framework that integrates drift detection and Transfer Learning for real-time monitoring of con- tainerized workloads. By focusing on operational metrics, including CPU, memory, I/O, and network usage, rather than solely application-level features, the system de- tects early performance interference signals. A key innovation is the application of the Kolmogorov-Smirnov (KS) test to identify statistically significant shifts in these resource distributions and automatically differentiate between benign updates and potentially disruptive anomalies. When the drift threshold is exceeded, selective re- training is initiated, ensuring that the models maintain accuracy without incurring the overhead of indiscriminate or excessively frequent retraining.
Experimental validation employs Acme Air and the DeathStarBench Social Network, two diverse microservice platforms that exhibit varied resource consumption patterns. By systematically introducing new functionalities and monitoring workload evolution, the framework demonstrates how KS-based drift detection surpasses conventional static-threshold methods in identifying early-stage noisy neighbor scenarios. Transfer Learning preserves prior knowledge while adapting rapidly to novel resource-usage profiles, offering a cost-effective approach for continuous, high-precision anomaly detection.
In conclusion, this thesis bridges data drift detection and system-level performance monitoring within an MLOps context. It presents a scalable and proactive strategy for mitigating performance interference in multi-tenant, cloud-native environments, ultimately enhancing reliability and preserving service-level objectives
Helical Products Parts Display Stand
Full text linkHelical Products needs a new product display stand that will demonstrate their custom springs, couplings, and u-joints. Their current display stand, a marble run contraption, is large, loud, and difficult to transport, so they would like the new stand to remedy these issues. The goal of this senior project is to create a display stand that captures the attention of their customers with dynamic product demonstrations while addressing the issues outlined above.
The new display stand is also a marble run contraption, but instead of being almost 4 feet tall and 3 feet long, it is 23” tall, 26.25” long, and 11.75” deep. Additionally, instead of requiring transportation in 3 separate parts, it can be transported as a whole assembly, weighing about 39 pounds. In the final design, the display stand can be divided into ten major subassemblies that each focus on a major component. Six of these subassemblies are for Helical Products’ custom parts, which are the lateral bending spring, compression spring, torsion spring, parallel misalignment coupling, u-joints, and the wrap-spring clutch. Two of the subassemblies are designated for the marble track, and the remaining two subassemblies are for the motor housing and the stand’s frame. The majority of parts in this stand are custom-designed 3D printed parts made of PETG, while other parts include necessary secondary components such as standard couplings, shafts, bearings, and mounting plates. For a single display stand, the estimated material cost is about 1700
Optimizing and Training an SVM-Based Breast Cancer Tumor Classifier
Full text linkWith advancements in technology, turning to machine learning has become a popular choice for aiding clinicians in the diagnoses of breast cancer malignancies. While the neural networking approach has been vetted thoroughly, this work aims to take advantage of traditional machine learning techniques; mainly support vector machine learning and the optimizing of feature extraction. The discrete-wavelet transform is used in the feature extraction stage of machine learning. Previous works that use this feature extraction technique are analyzed and expanded upon by utilizing a variety of different wavelets as well as other color-spaces with the goal of achieving higher result metrics and efficiency. In addition to the Haar wavelet, the Coiflet, Daubechies, and Symlet wavelets are analyzed due to their potential ability to distinguish more distinct features in an image. Furthermore, different approaches of class separation are explored with the aim of being able to classify a single subclass of a malignant tumor successfully. By utilizing the hue layer of the HSV color-space, higher result metrics are achieved by using a third of the training data. By using a third of the input features, this results in tripling the efficiency of the feature extraction stage. These results are compared to the results of neural networking approaches found in literature
Son Care Foundation - Pergola Preconstruction
No full textThis paper discusses the design and preconstruction aspects of a pergola constructed at Thousand Hills Pet Resort, supporting a non-profit organization called the Son Care Foundation. Founded in 1994 by Jack Gould, the foundation is committed to fostering hope through various programs including mentorship for students, service dog training for PTSD victims, basic pet training, vehicle repair and turnaround, and dog daycare and lodging. This project serves Son Care Foundation’s dog daycare and lodging program, called Thousand Hills Pet Resort. The program was established in 2005, and all profits are used to fund the other programs operated by the Son Care Foundation. Thousand Hills Pet Resort provides multiple services such as boarding, daycare, grooming, and dog training. The pergola built on the property required a large amount of design work, in which I was supported by two architectural engineers. The preconstruction aspects to be discussed include material procurement, budgeting, scheduling, and the challenge of handling multiple design changes
Cal Poly Leaning Pine Arboretum Gazebo
No full textLocated on the north end of California Polytechnic State University’s campus, the Leaning Pine Arboretum gazebo stands as a place of gathering, relaxation, and functionality. Taking the place of a smaller existing structure, this project transforms the area, creating new opportunities and laying a foundation for future growth over the coming years. Along the way, several challenges arose, including design complications, permitting issues, and time constraints. Through this process, it became clear that obstacles were inevitable, and overcoming them taught valuable lessons in teamwork, coordination between multiple parties, relationship building, and project management. From September to December 2024, work continued on the project following completion of the foundation and framing by a prior team member. The next phase involved installing the decking material, demolishing, and reconstructing a section of the deck, and conducting a geographical survey to support further site development. This experience led to a more practical and in depth understanding of planning, budgeting, and addressing the needs and vision of a client. Not only has this gazebo physically transformed the space, but its seamless redwood design complements with the surrounding arboretum, creating a striking and sustainable centerpiece to be enjoyed by all who visit
Hawthorne Elementary Stage Set Construction
No full textThe Pirate Ship and Playhouse project, undertaken by California Polytechnic State University, San Luis Obispo Cal Poly Construction Management students, aimed to support arts education at Hawthorne Elementary School through the construction of theatrical set pieces for a Peter Pan performance. Initiated by Phil Barlow and guided by Anna Brennen from the school’s staff, the project integrated community service with practical learning in construction management. From initial design and material sourcing to final installation, the team emphasized safety, durability, and aesthetic appeal. Funding and logistical support from local businesses and Cal Poly facilitated the project’s success. This hands-on endeavor not only showcased technical skills in construction but also underscored the program’s commitment to real-world application and community engagement. Lessons learned included effective communication, adaptability in project management, and the transformative impact of construction on educational environments. Ultimately, the Pirate Ship and Playhouse project exemplified Cal Poly’s Learn by Doing philosophy, leaving a lasting positive impact on the school community
Drug Delivery, Disulfide Crosslinked Hydrogel Coating For Polydimethylsiloxane (PDMS) Implant Materials
No full textThough polydimethylsiloxane (PDMS) is often used in medical devices due to its desirable mechanical properties and biocompatibility, its surface is often prone to bacterial adhesion. This is especially a problem for catheters, where bacteria can adhere and form biofilms that can in turn result in Urinary Tract Infections (UTI). In order to deter the presence of UTI causing bacteria on PDMS catheters, methods were developed to formulate a disulfide crosslinked hydrogel coating and create a methodology to allow adhesion to a PDMS substrate. The hydrogel coating discussed in this paper will be synthesized by chemically crosslinking stimuli-responsive vesicles, that carry therapeutic agents, within the hydrogel matrix to allow for controlled release of a therapeutic at a target condition. The hydrogel formulation will also contain therapeutics outside of the vesicles, but inside the matrix to further allow for a passive release of antibiotics. To allow for strong adhesion between the coating and its PDMS substrate, the PDMS substrate will also be functionalized to engage in chemical crosslinking with the hydrogel coating. The result should be a stimuli responsive hydrogel coating that allows for passive diffusion and stimuli responsive release of antibiotics to inhibit biofilm formation