Inspiration

What it does

Accelify is a web application that recommends ServiceNow Technical Accelerators to companies with a trained, self-adaptive PyTorch model.

How we built it

We generated a dataset using the provided company and product information as the input and calculated the output by relating products to each other and to the industries they were used in. We designed and built a PyTorch model that could handle a variable number of products for each company and updates parameters as the number of ServiceNow's products increases.

Challenges we ran into

A challenge we ran into with machine learning was how to handle an increase in the number of products as the model's output was shaped around the number of products so when there was an increase, we designed the model to modify its architecture while retaining the learned parameters.

Accomplishments that we're proud of

An accomplishment we're proud of is designing and implementing a dataset and PyTorch model for a recommendation system where the model has a variable length of input and output.

What we learned

We learned how to transfer model parameters between PyTorch model layers and use embedding and padding to handle variable input.

What's next for Accelify

Share this project:

Updates