Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

EBV: Electronic Bee-Veterinarian for Principled Mining and Forecasting of Honeybee Time Series

Mst. Shamima Hossain, Christos Faloutsos, Boris Baer, Hyoseung Kim, Vassilis J. Tsotras

TL;DR

The EBV (Electronic Bee-Veterinarian) method is applied to multiple real-world time sequences, and it is found that it yields accurate forecasting, accurate forecasting (up to 49% improvement in RMSE compared to baselines), and segmentation.

Abstract

Honeybees are vital for pollination and food production. Among many factors, extreme temperature (e.g., due to climate change) is particularly dangerous for bee health. Anticipating such extremities would allow beekeepers to take early preventive action. Thus, given sensor (temperature) time series data from beehives, how can we find patterns and do forecasting? Forecasting is crucial as it helps spot unexpected behavior and thus issue warnings to the beekeepers. In that case, what are the right models for forecasting? ARIMA, RNNs, or something else? We propose the EBV (Electronic Bee-Veterinarian) method, which has the following desirable properties: (i) principled: it is based on a) diffusion equations from physics and b) control theory for feedback-loop controllers; (ii) effective: it works well on multiple, real-world time sequences, (iii) explainable: it needs only a handful of parameters (e.g., bee strength) that beekeepers can easily understand and trust, and (iv) scalable: it performs linearly in time. We applied our method to multiple real-world time sequences, and found that it yields accurate forecasting (up to 49% improvement in RMSE compared to baselines), and segmentation. Specifically, discontinuities detected by EBV mostly coincide with domain expert's opinions, showcasing our approach's potential and practical feasibility. Moreover, EBV is scalable and fast, taking about 20 minutes on a stock laptop for reconstructing two months of sensor data.

EBV: Electronic Bee-Veterinarian for Principled Mining and Forecasting of Honeybee Time Series

TL;DR

The EBV (Electronic Bee-Veterinarian) method is applied to multiple real-world time sequences, and it is found that it yields accurate forecasting, accurate forecasting (up to 49% improvement in RMSE compared to baselines), and segmentation.

Abstract

Honeybees are vital for pollination and food production. Among many factors, extreme temperature (e.g., due to climate change) is particularly dangerous for bee health. Anticipating such extremities would allow beekeepers to take early preventive action. Thus, given sensor (temperature) time series data from beehives, how can we find patterns and do forecasting? Forecasting is crucial as it helps spot unexpected behavior and thus issue warnings to the beekeepers. In that case, what are the right models for forecasting? ARIMA, RNNs, or something else? We propose the EBV (Electronic Bee-Veterinarian) method, which has the following desirable properties: (i) principled: it is based on a) diffusion equations from physics and b) control theory for feedback-loop controllers; (ii) effective: it works well on multiple, real-world time sequences, (iii) explainable: it needs only a handful of parameters (e.g., bee strength) that beekeepers can easily understand and trust, and (iv) scalable: it performs linearly in time. We applied our method to multiple real-world time sequences, and found that it yields accurate forecasting (up to 49% improvement in RMSE compared to baselines), and segmentation. Specifically, discontinuities detected by EBV mostly coincide with domain expert's opinions, showcasing our approach's potential and practical feasibility. Moreover, EBV is scalable and fast, taking about 20 minutes on a stock laptop for reconstructing two months of sensor data.
Paper Structure (16 sections, 3 theorems, 6 equations, 8 figures, 2 tables)

This paper contains 16 sections, 3 theorems, 6 equations, 8 figures, 2 tables.

Table of Contents

  1. Motivation & Background
  2. Method & Technical Solution
  3. Related Work
  4. DataSet and Experimental Setup
  5. Proposed Method: EBV
  6. Proposed Model: EBVmodel.
  7. Justification
  8. Proposed Algorithm: EBVfit&cut.
  9. Empirical Evaluation
  10. Q1 - Effectiveness: Event-detection.
  11. Q1 - Effectiveness: Forecasting
  12. Q2 - Explainability
  13. Q3 - Scalability
  14. Q4 - Observations
  15. Conclusions, Significance and Impact
  16. ...and 1 more sections

Key Result

Lemma 2.1

Given the external temperature, ${\Theta_{ext}(t)} \xspace$, and the strengths ${s_h} \xspace$ and ${s_c} \xspace$ of honeybees for warming up and cooling down the hive, respectively, the hive core temperature, ${\Theta(t)} \xspace$, would obey:

Figures (8)

  • Figure 1: EBV at work: (a) deployed sensors (b) reconstruction and event detection (c) EBV wins on forecasting.
  • Figure 2: EBVfit&cut
  • Figure 3: EBV is effective & explainable: Accurate segments: EBVfit&cut fits (red line) the data (blue line) well and finds cut-points (brown vertical line) followed by unexpected events (red boxes: intensive hive inspection, sensor recharge and replace). Notice that the baselines are not applicable (cannot do segmentation).
  • Figure 4: EBV forecasts well: The seven days ahead forecasting (dotted red line) is close to the actual core temperature (solid blue line).
  • Figure 5: EBV wins against baseline in terms of forecasting accuracy (up to 49% improvement). Error bars show 1 standard deviation.
  • ...and 3 more figures

Theorems & Definitions (3)

  • Lemma 2.1
  • Lemma 3.1
  • Lemma 3.2