Irregular Sensor Data Compression

← Back to all benchmarks

Date: 2024-05-01

Name: Irregular Sensor Data Compression

Domain: High Energy Physics

Focus: Real-time compression of sparse sensor data with autoencoders

Task Types: Compression

Metrics: MSE, Compression ratio

Models: Autoencoder, Quantized autoencoder

AI/ML Motif: Generative

Resources

Benchmark: Visit

Datasets: Custom synthetic irregular sensor dataset

Results: Gemini LLM Deep Research

Keywords

compression autoencoder sparse data irregular sampling

Citation

• Javier Duarte, Nhan Tran, Ben Hawks, Christian Herwig, Jules Muhizi, Shvetank Prakash, and Vijay Janapa Reddi. Fastml science benchmarks: accelerating real-time scientific edge machine learning. 2022. URL: https://arxiv.org/abs/2207.07958, arXiv:2207.07958.

@misc{duarte2022fastmlsciencebenchmarksaccelerating2,
  archiveprefix = {arXiv},
  author        = {Javier Duarte and Nhan Tran and Ben Hawks and Christian Herwig and Jules Muhizi and Shvetank Prakash and Reddi, Vijay Janapa},
  eprint        = {2207.07958},
  primaryclass  = {cs.LG},
  title         = {FastML Science Benchmarks: Accelerating Real-Time Scientific Edge Machine Learning},
  url           = {https://arxiv.org/abs/2207.07958},
  year          = {2022}
}

Ratings

CategoryRating

Software

3.00

Not containerized; Full automation and documentation could be improved

Specification

4.00

Exact latency or resource constraints not numerically specified

Dataset

5.00

All criteria met

Metrics

5.00

All criteria met

Reference Solution

4.00

Not fully documented or automated for reproducibility

Documentation

4.00

Setup for deployment (e.g., FPGA pipeline) requires familiarity with tooling

Average rating: 4.17/5

Radar plot

Edit: edit this entry