Research
at Ikigai

Automating the pre-task of most data-driven analysis is that of “stitching” multiple data sources together.

DeepMatch learns the relationship between columns of a dataset and then uses it to further match the rows
‍
Our customers have used DeepMatch to do SKU mapping and new product launch in apparel retail, auditing in insurance, compliance in financial market, entity resolution across verticals and more

Automating the task of prediction - filling missing values in a spreadsheet.

DeepMatch learns the relationship between columns of a dataset and then uses it to further match the rows
‍
Our customers have used DeepMatch to do SKU mapping and new product launch in apparel retail, auditing in insurance, compliance in financial market, entity resolution across verticals and more

Running scenario analysis on steroids to make recommendations using reinforcement learning. 

Using scalable reinforcement learning computational infrastructure in addition to its computational technology, DeepPlan can compute scenario analysis for 1019 scenarios. 

Customers have used DeepPlan to create and find the optimal policies for all types of optimization problems, including labor forecasting

In terms of representation power, graphical models are at least as powerful, if not more, compared to the traditional neural networks – in effect, neural networks are deterministic versions of graphical models. That is, graphical models are neural networks on steroids.

In an MIT patent, we outline the architecture to explain how it enables computation to scale with data. In effect, it brings computation to data rather than the traditional approach of bringing data to compute.  The ability to compute with LGM enables >13x faster scientific computation on a commodity laptop compared to modern infrastructure using 68 parallel machines!