This is the repository for the Python PH talk about Deep Learning for Unstructured Data: Application on Image Search
I was fortunate enough to be able to share my expertise with Deep Learning at one of the PythonPH's meetups in partnership with Kalibrr.
I talked about how we can leverage Python and its rich scientific and data analysis toolkits to create a working image search engine. I also emphasized the importance of abstract vector spaces and the general technique of joint training of different models to merge two different vector spaces.
The link above presents an overview on machine learning, deep learning, unstructured data, and how Python as a programming language serves as a universal tool for applying deep learning and data processing to real-world problems.
The link to the demo contains technical details in building a working protoype for an Image Search Engine with unsupervised image-title-caption dataset scraped from the internet.