幽灵UMAP2:测量和分析 (r,d) -UMAP的稳定性
IEEE transactions on visualization and computer graphics
|December 4, 2025
概括
统一的多重近似和投影 (UMAP) 的结果可能是不稳定的,因为它的随机过程. 我们引入 (r,d) 稳定性来量化和分析投影稳定性,改进UMAP.
科学领域:
- 数据科学数据科学数据科学
- 机器学习 机器学习
- 减小尺寸性的减小方法
背景情况:
- 统一的多重近似和投影 (UMAP) 是一种流行的缩小维度的技术.
- 优化UMAP的随机性质可能导致投影结果不稳定.
- 这种随机性对UMAP可靠性的影响尚不清楚.
研究的目的:
- 引入一个新的框架, (r,d) 稳定性,用于分析UMAP预测的稳定性.
- 量化随机因素,如初始位置和负采样,对UMAP结果的影响.
- 为评估和改善UMAP投影稳定性提供工具和指导方针.
主要方法:
- 引入"幽灵" (重复的数据点) 来表示因随机性而导致的潜在位置变化.
- (r,d) 稳定性的定义是基于将幽灵投射限制在特定半径内.
- 开发一种适应性投降方案,以高效计算幽灵投影.
主要成果:
- 提出的 (r,d) 稳定性框架有效地分析了UMAP中随机性的影响.
- 适应性投降方案可将运行时间缩短高达60%,同时保留约90%的不稳定点.
- 一个可视化工具有助于对数据点投影稳定性的交互探索.
结论:
- (r,d) 稳定性框架为评估UMAP预测可靠性提供了一个强大的方法.
- 适应性放弃方案提供了显著的计算效率增长.
- 开发的工具和指导方针有助于研究人员获得更稳定和可解释的UMAP结果.
相关概念视频
Response Surface Methodology
583
Response Surface Methodology (RSM) is a collection of statistical and mathematical techniques used to develop, improve, and optimize processes. It is particularly valuable when many input variables or factors potentially influence a response variable.
The process of RSM involves several key steps:
The process of RSM involves several key steps:
583
The R Chart
358
In statistical process control, control charts, particularly R charts, are instrumental in monitoring process variations and identifying non-random patterns that run charts might miss. R charts track the variability within process subgroups, which is crucial when standard deviation use is impractical or unknown process variations exist.
R charts are pivotal for pinpointing shifts in process variability. Stability is indicated when all data points remain within the defined upper and lower...
R charts are pivotal for pinpointing shifts in process variability. Stability is indicated when all data points remain within the defined upper and lower...
358
Drug Product Stability
217
The long-term stability of drug products is critical to ensuring their quality, safety, and effectiveness over time. Stability directly influences a product's ability to maintain its intended characteristics, ensuring it performs as expected during its intended shelf life. Key attributes such as drug potency, impurities, dissolution, and other physicochemical measures of performance are tested to assess stability. These parameters indicate how well the product retains its quality over time and...
217
Microtubule Instability
6.0K
Microtubules are hollow cylindrical filaments having a diameter of approximately 25 nm and a length that varies from 200 nm to 25 μm. GTP-bound tubulin subunits form αβ-heterodimers for microtubule assembly. These core building blocks interact longitudinally, polymerizing into protofilaments. The protofilaments then interact with one another through lateral bonding forces to form stable cylindrical microtubules. These cylindrical filaments are dynamic as they undergo repeated...
6.0K
RNA Stability
35.6K
Intact DNA strands can be found in fossils, while scientists sometimes struggle to keep RNA intact under laboratory conditions. The structural variations between RNA and DNA underlie the differences in their stability and longevity. Because DNA is double-stranded, it is inherently more stable. The single-stranded structure of RNA is less stable but also more flexible and can form weak internal bonds. Additionally, most RNAs in the cell are relatively short, while DNA can be up to 250 million...
35.6K
Interpreting R Charts
314
R chart, or range chart, is a fundamental tool in statistical process control used to monitor the variability within a process. It complements the X-bar (x̄) chart by focusing on the range of the data, rather than individual values, providing a clear picture of the process dispersion over time.
An R chart plots the range of subsets of measurements collected from a process. Each point on the chart represents the range—defined as the difference between the maximum and minimum...
An R chart plots the range of subsets of measurements collected from a process. Each point on the chart represents the range—defined as the difference between the maximum and minimum...
314


