| 1. |
- Russo, Stefania, et al.
(författare)
-
The value of human data annotation for machine learning based anomaly detection in environmental systems
- 2021
-
Ingår i: Water Research. - : Elsevier BV. - 0043-1354. ; 206
-
Tidskriftsartikel (refereegranskat)abstract
- Anomaly detection is the process of identifying unexpected data samples in datasets. Automated anomaly detection is either performed using supervised machine learning models, which require a labelled dataset for their calibration, or unsupervised models, which do not require labels. While academic research has produced a vast array of tools and machine learning models for automated anomaly detection, the research community focused on environmental systems still lacks a comparative analysis that is simultaneously comprehensive, objective, and systematic. This knowledge gap is addressed for the first time in this study, where 15 different supervised and unsupervised anomaly detection models are evaluated on 5 different environmental datasets from engineered and natural aquatic systems. To this end, anomaly detection performance, labelling efforts, as well as the impact of model and algorithm tuning are taken into account. As a result, our analysis reveals the relative strengths and weaknesses of the different approaches in an objective manner without bias for any particular paradigm in machine learning. Most importantly, our results show that expert-based data annotation is extremely valuable for anomaly detection based on machine learning.
|
|
| 2. |
- Sharma, Sapna, et al.
(författare)
-
Loss of Ice Cover, Shifting Phenology, and More Extreme Events in Northern Hemisphere Lakes
- 2021
-
Ingår i: Journal of Geophysical Research - Biogeosciences. - : American Geophysical Union (AGU). - 2169-8953 .- 2169-8961. ; 126:10
-
Tidskriftsartikel (refereegranskat)abstract
- Long-term lake ice phenological records from around the Northern Hemisphere provide unique sensitive indicators of climatic variations, even prior to the existence of physical meteorological measurement stations. Here, we updated ice phenology records for 60 lakes with time-series ranging from 107-204 years to provide the first re-assessment of Northern Hemispheric ice trends since 2004 by adding 15 additional years of ice phenology records and 40 lakes to our study. We found that, on average, ice-on was 11.0 days later, ice-off was 6.8 days earlier, and ice duration was 17.0 days shorter per century over the entire record for each lake. Trends in ice-on and ice duration were six times faster in the last 25-year period (1992-2016) than previous quarter centuries. More extreme events in recent decades, including late ice-on, early ice-off, shorter periods of ice cover, or no ice cover at all, contribute to the increasing rate of lake ice loss. Reductions in greenhouse gas emissions could limit increases in air temperature and abate losses in lake ice cover that would subsequently limit ecological, cultural, and socioeconomic consequences, such as increased evaporation rates, warmer water temperatures, degraded water quality, and the formation of toxic algal blooms.
|
|
