Preventive Detection of Mosquito Populations using Embedded Machine Learning on Low Power IoT Platforms
Bibliographic record
Abstract
We can accurately detect mosquito species with 80% accuracy using frequency spectrum analysis of insect wing-beat patterns when mapped to low-power embedded/IoT hardware. We combine energy-efficient hardware acceleration optimizations with algorithmic tuning of signal processing and machine-learning routines to deliver a platform for insect classification. The use of low power accelerator blocks in cheap embedded boards such as the Raspberry Pi 3 and Intel Edison, along with performance tuning of the software implementations enable a competitive implementation of mosquito classification task on standard datasets. Our approach demonstrates a concrete application of embedding intelligence in edge devices for reducing system-level energy needs instead of simply uploading sensory data directly to the cloud for post-processing. For the mosquito classification task, we are able to deliver classification accuracies as high as 80% with Intel Edison processing times as low as 5 ms per set of 8K audio samples and an energy use of 5 mJ per sample (2 months of continuous non-stop use on an AA battery with 2000 mAh capacity or longer depending on insect activity). We envision a network of connected sensors and embedded/IoT platforms deployed in vulnerable such as construction sites, mines, areas of known mosquito activity, ponds, riverfronts, or other areas with standing water bodies. In our experiments, targeting a 20% packet loss rate, we observed the ad-hoc WiFi range for mesh networks using the Raspberry Pi 3 boards to be 14 m while the Photon board connecting to infrastructure WiFi router nodes can stretch this to 35 m.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".