MIT’s Sarah Alnegheimish Develops Orion: An Open-Source Anomaly Detection Framework for All
In the realm of machine learning, accessibility, transparency, and trustworthiness are paramount. Sarah Alnegheimish, a PhD student in Principal Research Scientist Kalyan Veeramachaneni’s Data-to-AI group at MIT’s Laboratory for Information and Decision Systems (LIDS), is making significant strides in these areas. Alnegheimish focuses her energy on developing Orion, an open-source machine learning framework and time series library designed to detect anomalies without supervision in large-scale industrial and operational settings.
Alnegheimish’s upbringing instilled in her a deep appreciation for shared knowledge. “I think growing up in a home where education was highly valued is part of why I want to make machine learning tools accessible.” Her early exposure to MIT OpenCourseWare further solidified her belief in accessible education. “I learned to view accessibility as the key to adoption. To strive for impact, new technology needs to be accessed and assessed by those who need it. That’s the whole purpose of doing open-source development.”
Orion addresses the critical need for anomaly detection in various sectors. Unusual patterns in network traffic can signal cybersecurity threats, abnormal sensor readings in machinery can predict failures, and monitoring vital signs can improve patient care. Orion utilizes statistical and machine learning models that are continuously logged and maintained, offering an end-to-end program for analyzing signals, comparing anomaly detection methods, and investigating anomalies.
“With open source, accessibility and transparency are directly achieved. You have unrestricted access to the code, where you can investigate how the model works through understanding the code. We have increased transparency with Orion: We label every step in the model and present it to the user.” This transparency fosters trust in the model’s reliability.
Alnegheimish is exploring innovative ways to leverage pre-trained models for anomaly detection. “When I first started my research, all machine-learning models needed to be trained from scratch on your data. Now we’re in a time where we can use pre-trained models,” she says. By repurposing models trained for forecasting, Alnegheimish aims to reduce the time and computational costs associated with anomaly detection. She enables them to detect anomalies through prompt-engineering, without any additional training.
Accessibility is at the core of Orion’s design. Alnegheimish emphasizes the importance of system development alongside model development. “Before I came to MIT, I used to think that the crucial part of research was to develop the machine learning model itself or improve on its current state. With time, I realized that the only way you can make your research accessible and adaptable for others is to develop systems that make them accessible. During my graduate studies, I’ve taken the approach of developing my models and systems in tandem.” By creating universal representations for all models with simplified components, she ensures that users can easily adapt and implement the framework.
Alnegheimish’s work extends to making AI more approachable through LLM. The LLM agent implemented is able to connect to Orion without users needing to know the small details of how Orion works. For her software, users only know two commands: Fit and Detect. Fit allows users to train their model, while Detect enables them to detect anomalies.
Orion has garnered significant attention, with over 120,000 downloads and a thousand users marking it as a favorite on Github. “Traditionally, you used to measure the impact of research through citations and paper publications. Now you get real-time adoption through open source.” Sarah Alnegheimish’s work underscores the transformative potential of open-source frameworks in democratizing access to advanced machine learning tools.
