THE POLYCHORD CORE TOOL. CAN FIT MODELS TO DATA IN HIGH DIMENSIONAL SCENARIOS, WHERE THE PROBLEM IS COMPLEX AND THERE ARE MANY VARIABLES. THIS IS ITS KEY ADVANTAGE OVER ANY OTHER DATA TOOL.
PolyChord is a spinout from Cambridge University, formed in 2017 to commercialise the unique and cutting-edge analytical data tool developed by two cosmology professors (Professor Mike Hobson & Professor Anthony Lasenby) both now directors of the company. The data they had to deal with came from the Planck satellite and went from the present day to the Big Bang, spanning ten billion years. The tool, PolyChord, successfully extracted the desired information from that data – and we now look to apply it to commercial instances where there are similar Big Data problems with issues of high dimensionality, many variables, complexity, and dirtiness. Many people these days claim to have smart data tools – PolyChord is a genuine step change ahead of the rest and can fit models to data when the task is exceptionally challenging, in order to extract information and make predictions.
We have made progress by collaborating with large, dynamic successful companies, who have gathered data, prepared it for analysis, but have then run into difficulties with the number of variables, and high dimensionality. Fitting traditional models to extract this information is then impossible – this is where our PolyChord tool really delivers, and global optimisation is also achievable in many circumstances. PolyChord has this year been used to optimise neural networks, where we treat the honeycomb-like surface of the neural network as a high dimensional data problem in its own right and solve it using PolyChord. The result is PolyNet, a second tool for the company, and a new way of using neural networks for more accurate Machine Learning.
For more information on current and upcoming projects, visit our PolyChord Blog.
Please tell us some details about the nature of your enquiry