- Award ID(s):
- 1852002
- PAR ID:
- 10323864
- Date Published:
- Journal Name:
- Sensors
- Volume:
- 21
- Issue:
- 18
- ISSN:
- 1424-8220
- Page Range / eLocation ID:
- 6052
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
null (Ed.)Green homes require informed energy management decisions. For instance, it is preferable that a comfortable internal temperature is achieved through natural, energy-efficient means such as opening doors or lowering shades as opposed to turning on the air conditioning. This requires the control agent to understand the complex system dynamics of the home: will opening the window raise or lower the temperature in this particular situation? Unfortunately, developing mathematical models of a suburban home situated in its natural environment is a significant challenge, while performing real-world experiments is costly, takes a long time and depends on external circumstances beyond the control of the experimenter. In this paper, we describe the architecture of a physical, small scale model of a suburban home and its immediate exterior environment. Specific scenarios can be enacted using Internet of Things (IoT) actuators that control the doors and windows. We use a suite of IoT sensors to collect data during the scenario. We use deep learning-based temporal regression models to make predictions about the impact of specific actions on the temperature and humidity in the home.more » « less
-
Whether to evacuate a nursing home (NH) or shelter in place in response to the approaching hurricane is one of the most complex and difficult decisions encountered by nursing home administrators. A variety of factors may affect the evacuation decision, including storm and environmental conditions, nursing home characteristics, and the dwelling residents’ health conditions. Successful prediction of evacuation decision is essential to proactively prepare and manage resources to meet the surge in nursing home evacuation demands. In current nursing home emergency preparedness literature, there is a lack of analytical models and studies for nursing home evacuation demand prediction. In this paper, we propose a predictive analytics framework by applying machine learning techniques, integrated with domain knowledge in NH evacuation research, to extract, identify and quantify the effects of relevant factors on NH evacuation from heterogeneous data sources. In particular, storm features are extracted from Geographic Information System (GIS) data to strengthen the prediction accuracy. To further illustrate the proposed work and demonstrate its practical validity, a real-world case study is given to investigate nursing home evacuation in response to recent Hurricane Irma in Florida. The prediction performance among different predictive models are also compared comprehensively.more » « less
-
Abstract Fractionally doped perovskites oxides (FDPOs) have demonstrated ubiquitous applications such as energy conversion, storage and harvesting, catalysis, sensor, superconductor, ferroelectric, piezoelectric, magnetic, and luminescence. Hence, an accurate, cost-effective, and easy-to-use methodology to discover new compositions is much needed. Here, we developed a function-confined machine learning methodology to discover new FDPOs with high prediction accuracy from limited experimental data. By focusing on a specific application, namely solar thermochemical hydrogen production, we collected 632 training data and defined 21 desirable features. Our gradient boosting classifier model achieved a high prediction accuracy of 95.4% and a high F1 score of 0.921. Furthermore, when verified on additional 36 experimental data from existing literature, the model showed a prediction accuracy of 94.4%. With the help of this machine learning approach, we identified and synthesized 11 new FDPO compositions, 7 of which are relevant for solar thermochemical hydrogen production. We believe this confined machine learning methodology can be used to discover, from limited data, FDPOs with other specific application purposes.more » « less
-
A smart home with a controller that can understandand predict the interaction between the external environment and the user’s behavior and preferences can provide significant energy efficiency and savings. Unfortunately, experimentation of real world homes for the development of such a controller is prohibitively expensive. In this paper we describe techniques through which such experiments can be performed on scaled testbed with an accelerated time. We illustrate how the modeling of different geographical areas can be performed by the mapping of the model’s temperature and time to their real-world equivalents. We train three different machine learning models for predicting different sensor readings in the testbed, and find that the achieved predictive accuracy supports the feasibility of the development of future smart climate controllers.more » « less
-
Abstract Background Animal-borne sensors (‘bio-loggers’) can record a suite of kinematic and environmental data, which are used to elucidate animal ecophysiology and improve conservation efforts. Machine learning techniques are used for interpreting the large amounts of data recorded by bio-loggers, but there exists no common framework for comparing the different machine learning techniques in this domain. This makes it difficult to, for example, identify patterns in what works well for machine learning-based analysis of bio-logger data. It also makes it difficult to evaluate the effectiveness of novel methods developed by the machine learning community.
Methods To address this, we present the Bio-logger Ethogram Benchmark (BEBE), a collection of datasets with behavioral annotations, as well as a modeling task and evaluation metrics. BEBE is to date the largest, most taxonomically diverse, publicly available benchmark of this type, and includes 1654 h of data collected from 149 individuals across nine taxa. Using BEBE, we compare the performance of deep and classical machine learning methods for identifying animal behaviors based on bio-logger data. As an example usage of BEBE, we test an approach based on self-supervised learning. To apply this approach to animal behavior classification, we adapt a deep neural network pre-trained with 700,000 h of data collected from human wrist-worn accelerometers.
Results We find that deep neural networks out-perform the classical machine learning methods we tested across all nine datasets in BEBE. We additionally find that the approach based on self-supervised learning out-performs the alternatives we tested, especially in settings when there is a low amount of training data available.
Conclusions In light of these results, we are able to make concrete suggestions for designing studies that rely on machine learning to infer behavior from bio-logger data. Therefore, we expect that BEBE will be useful for making similar suggestions in the future, as additional hypotheses about machine learning techniques are tested. Datasets, models, and evaluation code are made publicly available at
https://github.com/earthspecies/BEBE , to enable community use of BEBE.