You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group


  • Fea­ture Engi­neer­ing: Data scientist’s Secret Sauce !
    • It is very tempt­ing for  data sci­ence prac­ti­tion­ers to opt for the best known  algo­rithms for a given problem.However It’s not the algo­rithm alone , which can pro­vide the best solu­tion  ; Model built on care­fully engi­neered and selected fea­tures can pro­vide far bet­ter results.
  • pix2code: Gen­er­at­ing Code from a Graph­i­cal User Inter­face Screen­shot
    • In this paper, we show that Deep Learn­ing tech­niques can be lever­aged to auto­mat­i­cally gen­er­ate code given a graph­i­cal user inter­face screen­shot as input. Our model is able to gen­er­ate code tar­get­ing three dif­fer­ent plat­forms (i.e. iOS, Android and web-based tech­nolo­gies) from a sin­gle input image with over 77% of accuracy.
    • As writ­ten in the paper, the datasets will be made avail­able but noth­ing is said about the source code. How­ever, because of the unex­pected amount of inter­est in this project, the pix2code imple­men­ta­tion described in the paper will also be open-sourced in this repo together with the datasets.
  • https://github.com/reiinakano/xcessiv
    • A web-based appli­ca­tion for quick and scal­able hyper­pa­ra­me­ter tun­ing and stacked ensem­bling in Python
  • Intro­duc­tion to Prob­a­bilis­tic Mod­el­ling and Machine Learning
  • A tuto­r­ial on how to deploy a Dock­erised deep learn­ing appli­ca­tion on Azure Con­tainer Ser­vices
    • Deploy­ing machine learn­ing mod­els can often be tricky due to their numer­ous depen­den­cies, deep learn­ing mod­els often even more so. One of the ways to over­come this is to use Docker con­tain­ers. Unfor­tu­nately, it is rarely straight-forward. In this tuto­r­ial, we will demon­strate how to deploy a pre-trained deep learn­ing model using Azure Con­tainer Ser­vices, which allows us to orches­trate a num­ber of con­tain­ers using DC/OS.
  • Cal­cu­lus Made Easy 2nd ed, 1914
  • Anti-Money Laun­der­ing and AI at HSBC
    Anti-Money Laundering and AI at HSBC
    Anti-Money Laun­der­ing and AI at HSBC
    • Anti-Money Laun­der­ing is a par­tic­u­larly chal­leng­ing area of reg­u­la­tion for banks and even more so for large, geo­graph­i­cally diverse insti­tu­tions. What makes AML such a dif­fi­cult prob­lem to solve is that it involves com­plex data, detailed work­flows, and sig­nif­i­cant human involve­ment.  In short it is per­fect for AI.