Machine Learning

Approaches to augmenting data and challenges with limited datasets

Data augmentation is a technique used to increase the amount of data available for machine learning models when faced with limited datasets. In the field of eCommerce, where R&D teams are developing new technologies to improve user experience, this is a common issue. With limited data available, training machine learning models to categorize product images can be difficult. This is where data augmentation techniques come in, allowing for the creation of new synthetic data from the existing dataset to improve the performance of the model. In this blog post, we will explore the different approaches to data augmentation and the challenges that arise when dealing with limited datasets in eCommerce.

How Does a Machine Find an Object? An introduction to Machine Learning Computer Vision

The difference between how machines “see” and identify objects compared to humans may not be so obvious. We are able to instantly identify any object we see and sometimes even objects we’ve never seen before, because we are using our memories and collective experiences. Without conscious thought, we create a complex web of connections between any new object we see and through the process of comparison we can approximate its purpose. Sometimes we even can correctly guess the name of this object. Even with incredibly fast GPU’s, improved algorithms and an unbelievably huge amount of collected data, machines are not yet able to do this. That’s because it’s still extremely difficult for machines to think outside of the box.