I recently watched this lecture by Dmitry Krotov and found it very interesting so I thought it would make a good paper to try to reproduce. My original thoughts were that this could potentially solve the adversarial vulnerability inherent in most modern convolutional networks. This blog post are the details of this reproduction and some […] Read more
Any machine learning practitioner will tell you that there is a certain point in time where trying to eke out more performance/accuracy from a project seems like more effort than its worth. At PicNet we always suggest customers run a 4-6-week proof of concept to get an idea of the potential accuracy for their project. […] Read more
I am told quite frequently that an organisation is not “mature” enough for machine learning. By this, managers usually mean one of the following things items are not at a level deemed adequate: The data warehousing infrastructure The data sophistication of users The consolidation of multiple data sources The long-term data and analytics strategy […] Read more
A very common problem I find in the industry is senior managers not fully understanding how best to utilize Machine Learning technologies to help their business. A common misconception is that these projects should be treated the same as traditional analytics and whilst they can serve that purpose I think that is leaving most of […] Read more
Introducing XGBoost.Net – .Net wrappers for the awesome XGBoost library XGBoost is a big part of our Machine Learning and Predictive Analytics toolkit here at PicNet. We use it almost heavily for our proof of concept and prototype work and it is always present in ensembles for production systems. We usually host our python models on […] Read more
[Edit]: It appears the XGBoost team has fixed pip builds on Windows. There are also nightly artifacts generated. As such, I hereby turn off my nightly builds. [Edit]: These builds (since 19th of Dec 2016) now have GPU support. If this causes any issues let me know and I’ll create 2 separate binaries. I will be […] Read more
Congratulations to the XGBoost team who have sorted out a lot of issues with XGBoost build on windows. This is how I managed to build XGBoost on my environment: Windows 10 Visual Studio 2015 Install CMake CMake can be downloaded from here: https://cmake.org/download/ I downloaded the 64 bit zip file: https://cmake.org/files/v3.6/cmake-3.6.2-win64-x64.zip and unzipped it to c:\dev\cmake-3.6.2-win64-x64. [Edit […] Read more
Requesting focus on an input field is surprisingly hard in angular2. Especially when the input field is inside a dialog or an *ngIf that is not initially visible. After much trial and error this is the solution I came up with. It works on my machine and my scenarios, no other guarantees provided: import {Directive, […] Read more
Getting Angular2 and IE9 working together is not very straight forward, since the move to release candidates some of the Shim libraries have been changed and others removed from the NG2 packages making things confusing. After some research this is what I have determined to be the minimum number of “workaround” libs required. Note I […] Read more
Upgrading to new the new form engine in Angular 2 (RC3) is fairly straight forward and not nearly as dawnting as some of the router changes in the past. So here is my quick and dirty how to which worked on my projects (no other guarantee). Firstly – Read the docs One of the issues […] Read more
0