The majority of modern consumer-level energy is generated by real-time smart
metering systems. These frequently contain anomalies, which prevent reliable
estimates of the series' evolution. This work introduces a hybrid modeling
approach combining statistics and a Convolutional Autoencoder with a dynamic
threshold. The threshold is determined based on Mahalanobis distance and moving
averages. It has been tested using real-life energy consumption data collected
from smart metering systems. The solution includes a real-time, meter-level
anomaly detection system that connects to an advanced monitoring system. This
makes a substantial contribution by detecting unusual data movements and
delivering an early warning. Early detection and subsequent troubleshooting can
financially benefit organizations and consumers and prevent disasters from
occurring.

创新的混合建模方法结合统计学和卷积自编码器，利用动态阈值（基于马氏距离和移动平均）检测实时智能计量系统中的异常，从而提供早期预警，以预防灾难并在经济上受益于组织和消费者。

基于动态阈值的卷积自编码器实时异常检测

A Real-time Anomaly Detection Using Convolutional Autoencoder with  Dynamic Threshold

Early warning for epilepsy patients is crucial for their safety and
well-being, in terms of preventing or minimizing the severity of seizures.
Through the patients' EEG data, we propose a meta learning framework for
improving prediction on early ictal signals. To better utilize the meta label
corrector method, we fuse the information from both the real data and the
augmented data from the latent Stochastic differential equation(SDE). Besides,
we also optimally select the latent dynamical system via distribution of
transition time between real data and that from the latent SDE. In this way,
the extracted tipping dynamical feature is also integrated into the meta
network to better label the noisy data. To validate our method, LSTM is
implemented as the baseline model. We conduct a series of experiments to
predict seizure in various long-term window from 1-2 seconds input data and
find surprisingly increment of prediction accuracy.

通过患者的脑电图数据，我们提出了一种基于元学习框架的早期癫痫信号预测方法，通过融合真实数据和隐式随机微分方程（SDE）生成的增广数据，以及优化选择潜在动力系统的转换时间分布，将提取的关键动力特征与元网络整合，以更好地标记嘈杂数据，并使用 LSTM 作为基准模型进行验证。我们进行了一系列实验，预测不同长期窗口（1-2 秒的输入数据），发现预测准确性有意想不到的增加。

基于保稳转折潜在随机动态系统和元标签修正的早期预警

Early Warning via tipping-preserving latent stochastic dynamical system  and meta label correcting

Earth is bombarded by meteors, occasionally by one large enough to cause a
significant explosion and possible loss of life. Although the odds of a deadly
asteroid strike in the next century are low, the most likely impact is by a
relatively small asteroid, and we suggest that the best mitigation strategy in
the near term is simply to move people out of the way. We describe an "early
warning" system that could provide a week's notice of most sizable asteroids or
comets on track to hit the Earth. This system, dubbed "Asteroid
Terrestrial-impact Last Alert System" (ATLAS), comprises two observatories
separated by about 100km that simultaneously scan the visible sky twice a
night, and can be implemented immediately for relatively low cost. The
sensitivity of ATLAS permits detection of 140m asteroids (100 Mton impact
energy) three weeks before impact, and 50m asteroids a week before arrival. An
ATLAS alarm, augmented by other observations, should result in a determination
of impact location and time that is accurate to a few kilometers and a few
seconds. In addition to detecting and warning of approaching asteroids, ATLAS
will continuously monitor the changing universe around us: most of the variable
stars in our galaxy, many micro-lensing events from stellar alignments,
luminous stars and novae in nearby galaxies, thousands of supernovae, nearly a
million quasars and active galactic nuclei, tens of millions of galaxies, and a
billion stars. With two views per day ATLAS will make the variable universe as
familiar to us as the sunrise and sunset.

我们描述了一个早期警报系统（ATLAS），可以提供大多数可撞击地球的中等大小的小行星或彗星的一周通知。该系统由两个天文台组成，可以立即实施，并具有相对较低的成本。