Object2Vec Algorithm is an Unsupervised Learning algorithm. The algorithm compares pairs of data points and preserves the semantics of the relationship between the pairs. The algorithm creates embeddings that can be used by other algorithms downstream. The embeddings are low-dimensional dense embeddings of high-dimensional objects. Object2Vec can be used for product search, item matching and customer profiling.
Customer profiling is used in personalization and recommendation systems. Object2Vec Algorithm can deliver predictions that are more accurate than manual methods based on customer attributes such as age or gender.
- AWS docs: https://docs.aws.amazon.com/sagemaker/latest/dg/object2vec.html
- AWS blog: https://aws.amazon.com/blogs/machine-learning/introduction-to-amazon-sagemaker-object2vec/
- Use case: https://blog.griddynamics.com/customer2vec-representation-learning-and-automl-for-customer-analytics-and-personalization/
|Data types and format||Tabular|
|Learning paradigm or domain||Unsupervised Learning|
|Problem type||Algorithm embeddings|
|Use case examples||Improve the data embeddings of the high-dimensional objects|
Object2Vec Algorithm training data is provided in Json lines format.
Model artifacts and inference
|Learning paradigm||Unsupervised Learning|
- Training is performed on a single instance which can be CPU or GPU. Multi-cored GPUs are also supported.
- GPU instances are used for inference.
AWS Partner Webinar: Object2Vec on Amazon SageMaker
This is a 50.57 minutes video by Kris Skrinak from AWS.