Huawei welcomes a selection of top-of-the-class students to their 2021 University Challenge!
This 4-week hackathon, hosted virtually, will gather data science students across the United Kingdom seeking to take their skills to the next level, through rigorous data-related challenges and win from the £15,000 prize pool!
Improving the digital world has been Huawei’s objective from the beginning, which is why finding the next generation of innovators will always be at the core of their mission.
How can we innovate and take data science to a whole new level?
This year's challenge will be: Indoor Positioning
Indoor positioning is the science used to locate devices in indoor environments where global navigation satellite systems (GNSS) aren't available. It is achieved using many different techniques that include the deployment of sensor networks, beacons, the exploitation of the wireless infrastructure, maps etc.
To get to the final level you must succeed in the first two, one being more challenging than the other! Don't get left behind.
Begin by tackling a geospatial data estimation challenge with access to labelled dataset. You will need to analyse the data and train model to predict how far two data points are from each other (indoor spaces).
In order to continue to the second level you must obtain a minimum score in the first task. Don't let it be game over for your team!
This challenge will account for 35% of the total score.
The second task will be a data classification/grouping challenge where teams will create clusters of data points represent physical areas from unlabelled data.
Only the top 6 teams will be invited to the finale. Be one of the few that advance to the final level!
This challenge will account for 65% of the total score.
You’ve aced all the previous levels! Congratulations, you’ve made it to the last one!
Selected number of the finalist teams will be invited to present their solution in front of Huawei’s jury at the virtual Award Ceremony.
You will be allowed to form teams of 2 or 3 participants (students only).
Teams will be able to submit their solutions once per day using the coding platform Isograd.
Judging will be based on precision and recall metrics with final scores calculated as a proportion of best performing algorithm (a score of 1.0) and the minimum accepted value (a score of 0).
Georeferenced data can be any type of information that is related to a specific position or area. This includes many different types of data, like population density, weather measurements, maps, traffic, census, satellite images, etc. The information is usually highly correlated to its position, therefore it can be used to extrapolate data from unknown areas, predict the position of unmapped measurements, and understand the events generating those measurements, relationships between features, among others.
Data science - artificial intelligence
Semi-supervised or unsupervised machine learning used as qualitative data mining to extract geospatial features from nosier data. Such data is commonly obtained with qualititave analysis and conventional data mining.
The ability to annotate location data at scale based on graph theory and similarity of data points. It is very common that Indoor Positioning Solutions employ radio-maps of Wi-Fi signals as reference points to compute smartphone positions. These reference points can form a knowledge graph considering their relation to the geospatial annotation and maps data.
Clustering & Classification
Spatial relationships and links between data points, stored as a geospatial graph, reduced into clusters then used to train classification function. Can be used for machine learning, data fusion, parameter optimization and many other algorithms to extract the information and to use it for positioning.
An Indoor Positioning Solutions (IPS) is a system that uses the geospatial data to estimate the position of a user or device in challenging scenarios (e.g. inside a train station). Many technologies like guided navigation, virtual or augmented reality, location sharing, business intelligence, security services and other technologies make use of this kind of system to provide location contextual information.
Indoor Positioning Systems not only show the location to the user, they also improve the experience of many other applications, boosting the location context data availability, and creating the foundation for many scientific, technical, and industrial applications.