Terms & Conditions | BMA Privacy Policy| Huawei Privacy Policy

The leading global provider of information and communications technology (ICT) infrastructure and smart devices, Huawei is thrilled to invite students from across Ireland to take part in this year’s European University Challenge 2020!

This three-week digital event will task participants with artificial intelligence and time-series anomaly detection.   

Theme:
Artificial Intelligence Challenge: Time Series Anomaly Detection

Time-series anomaly detection helps engineers to monitor the KPIs continuously and alert for potential incidents on time.

Participants will be given a list of time series KPI datasets. Each KPI CSV dataset contains timestamp, KPI value, request count, and anomaly label 4 columns. The anomaly label denotes whether a KPI value at a timestamp is an anomaly point. Participants can use these datasets to develop and test their algorithms to identify anomalies of KPI value as accurately as possible. 

Invited to participate are undergraduate, Master’s, and PhD students currently enrolled in an Irish University who may compete individually or as a team of up to three people. Collaborating on multiple teams is not permitted.

Undergraduate

Master's

PhD

FIRST PLACE

HUAWEI
MateBook 14

SECOND PLACE

HUAWEI P30 Pro

Or equivalent 

THIRD PLACE

HUAWEI WATCH
GT 2 Lake Cyan

Or equivalent 

Or equivalent 

In addition to individual prizes, all winners will be considered for internship positions!

Registration Opens

October 21, 2020
3:00pm

Opening Ceremony
& Challenge Begins

November 16, 2020

Registration Closes

December 7, 2020
11:59 pm

Final Submission
Deadline

December 8, 2020
11:59 pm

Finalist Selection

December 9, 2020 - December 13, 2020

Awards Ceremony

December 15, 2020

A new list of KPI datasets with an empty anomaly_label column will be provided for evaluation. Participants are required to fill the anomaly label column based on their algorithm outputs.

The accuracy of anomaly detection is calculated with standard precision, recall, and F1 score by comparing with the ground truth anomaly labelling which will not be provided to participants before ending the challenge.

For an incident contains a number of continuous anomaly points, first N (<10) will give double points in the accuracy calculation

• For detecting any anomaly of a timestamp, only data before the timestamp is allowed to be used for the input of the algorithm

• Manual algorithm parameter tuning are allowing for the whole dataset, but not allowed for each individual dataset.

Following the final submission deadline, a panel of judges will select the top 6 teams. These finalists will showcase their demos and presentations during the virtual closing ceremony at which time the judges will select and announce the winners. 

marie@bemyapp.com

ORGANIZED WITH

POWERED BY

© 2020. BEMYAPP, All rights reserved.

Terms & Conditions | BMA Privacy Policy| Huawei Privacy Policy