Graf Neural Networks (Gnns)

RECSYS: RAJEEV RASTOGI about three recommendation system challenges

by
RECSYS: RAJEEV RASTOGI about three recommendation system challenges

Rayev Rastogi, Vice President of Applied Science in Amazon’s International Emerging Stores Division. In a Keynote address of this year’s ACM Conference on Recommendation Systems (Recsys), which starts next week, Rajeev Rastogi, Vice President of Applied Science in Amazon’s International Emerging Stores Division, will discuss three problems that have been in its recommendation algorithms: Recommendations … Read more

Anomali detection for graph-based data

by
Anomali detection for graph-based data

Anomali detection is the identification of data that differs significantly from established norms, which may indicate damage activity. It is a particularly tough challenge in the case of graph -based data, where anomalid detection is not only based on data values, but on topological conditions with the graph. Becuse anomalies tend to be rare, it … Read more

Interpretable improvements of product recovery models

by
Interpretable improvements of product recovery models

The machine learning field is developing at a quick pace with the regular release of new models that promise improvements over their predecessors. However, evaluation of a new model for a particular use case is a time -consuming and resource -intensive process. It is a conundrum for online services such as Amazon’s store, which is … Read more

Benchmarking tools for graph-centered prediction modeling on databases

by
Benchmarking tools for graph-centered prediction modeling on databases

Relationship databases (RDBS) store huge amants of structured data across several interconnected tables. This rich relational information has great potential for predictable machine learning. However, the progress of prediction models on RDBs is currently lagging behind progress in other domains such as computer vision or natural-langage treatment. An important reason is the lack of established, … Read more