A5092355520- Distributed and Parallel Computing Systems
- Parallel Computing and Optimization Techniques
- Cloud Computing and Resource Management
- Handwritten Text Recognition Techniques
- ECG Monitoring and Analysis
- Geophysical Methods and Applications
- Topic Modeling
Google (United States)
2024
University of Washington
2024
Kernel task scheduling is important for application performance, adaptability to new hardware, and complex user requirements. However, developing, testing, debugging algorithms in Linux, the most widely used cloud operating system, slow difficult. We developed Enoki, a framework high velocity development of Linux kernel schedulers. Enoki schedulers are written safe Rust, system supports live upgrade policies into kernel, userspace debugging, bidirectional communication with applications. A...
Background/Aim: Smartwatch ECGs (SW-ECG) have emerged as a noninvasive solution to assess abnormal heart rhythms such atrial fibrillation (AF). However, SW-ECGs recorded in free-living settings are often noisy and yield Inconclusive results from automated AF detection algorithms. We developed deep learning model denoise single-lead reduce the number of inconclusive tracings that require manual overread. Methods: Using MIMIC-III waveform database (N=2,988 patients, 10.1% AF), we created pairs...
Kernel task scheduling is important for application performance, adaptability to new hardware, and complex user requirements. However, developing, testing, debugging algorithms in Linux, the most widely used cloud operating system, slow difficult. We developed Ekiben, a framework high velocity development of Linux kernel schedulers. Ekiben schedulers are written safe Rust, system supports live upgrade policies into kernel, userspace debugging, bidirectional communication with applications. A...
This paper introduces RETSim (Resilient and Efficient Text Similarity), a lightweight, multilingual deep learning model trained to produce robust metric embeddings for near-duplicate text retrieval, clustering, dataset deduplication tasks. We demonstrate that is significantly more accurate than MinHash neural embeddings, achieving new state-of-the-art performance on deduplication, adversarial retrieval benchmarks, spam clustering also introduce the W4NT3D benchmark (Wiki-40B 4dversarial...