Numenta Logo

The NAB Competition 2016

Taylor WirfsTaylor WirfsMarketing

Numenta Anomaly Benchmark (NAB) Competition

IEEE WCCI (World Congress on Computational Intelligence) 2016

We are excited to announce the Numenta Anomaly Benchmark (NAB) Competition! Last year, we released NAB, the first ever open-source benchmark for evaluating real-time anomaly detection algorithms. This year, we want to further expand NAB by including more real-world datasets and benchmarked algorithms, through a fun competition - with cash prizes.

The contest will be held in conjunction with the IEEE WCCI 2016 on July 25 – 29 in Vancouver, Canada. The competition offers two submission categories: Algorithms and Datasets. Want to see how your anomaly detection algorithm performs? Have a dataset with labeled anomalies? Then this contest is for you.

Entries can be submitted to either category, or both. All entries should be sent to

Numenta Anomaly Benchmark screenshot

Example entry of satellite data with labeled anomalies. See more examples.

Entry Categories

Algorithms Category

We are looking for algorithms that detect anomalies in streaming data.

Entries in this category must submit the following:

  • Detailed description of the algorithm and NAB results
  • Code to run the algorithm and re-create the results (GitHub link OK)
  • List of contributors and their contact information

We will evaluate your algorithm by running your code on NAB v1.0. Results will be added to the NAB repository scoreboard, following the conference.

Datasets Category

We are looking for real-world, time-series datasets with labeled anomalies.

Entries in this category must submit the following:

  • Data file(s) in CSV format
  • Anomaly labels, with timestamps at which the anomalies start
  • Detailed description of the data and anomalies
  • List of contributors and their contact information

We will evaluate the dataset on the following characteristics: types of anomalies, relevance to real applications, quality, difficulty for algorithms, quantity of data (more files in your set, the better), and feasibility for detection. Viable datasets will be added to the next version of NAB, shortly after the competition ends.

Prizes $$

All work and no play is no fun. We will be awarding up to $10,000 in cash prizes for your contributions to NAB.

Rules and Eligibility For a complete list of rules and contest eligibility, visit:

Deadline and Questions

Submissions are due by July 1, 2016. We encourage you to ask questions, and submit entries ASAP so we can ensure they meet all the requirements. Contact us with any questions at

Ready to get started? Check out these resources:

Taylor WirfsTaylor WirfsMarketing