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


  • Mul­ti­ple impu­ta­tion util­is­ing denois­ing autoen­coder for approx­i­mate Bayesian infer­ence
    • Miss­ing data is a wide­spread prob­lem in machine learn­ing. Bayesian infer­ence is a robust solu­tion to imput­ing miss­ing val­ues, par­tic­u­larly if mul­ti­ple impu­ta­tions are used to model the uncer­tainty regard­ing said values.”
  • Alibaba’s AI Out­guns Humans in Read­ing Test
    • Alibaba has devel­oped an arti­fi­cial intel­li­gence model that scored bet­ter than humans in a Stan­ford Uni­ver­sity read­ing and com­pre­hen­sion test.
  • Robo-Advisers Are Com­ing to Con­sult­ing and Cor­po­rate Strat­egy
    • Does a robot man­age your money? For many of us, the answer is yes. Online and algo­rith­mic invest­ment and finan­cial advice is easy to come by these days, usu­ally under the moniker of “robo-advisor.” Star­tups such as Wealth­front, Per­sonal Cap­i­tal, and Bet­ter­ment launched robo-advisors as indus­try dis­rup­tors, and incum­bents, such as Schwab’s (Intel­li­gent Advi­sor), Van­guard (Per­sonal Advi­sor Ser­vices), Mor­gan Stan­ley and Black­Rock have joined the fray with their own hybrid machine/advisor solutions.”
  • One model to learn them all
    • Sup­pose you ask me if I’d like any­thing to eat. I can say the word ‘banana’ (such that you hear it spo­ken), send you a text mes­sage whereby you see (and read) the word ‘banana,’ show you a pic­ture of a banana, and so on. All of these dif­fer­ent modal­i­ties (the sound waves, the writ­ten word, the visual image) tie back to the same con­cept – they are dif­fer­ent ways of ‘inputting’ the banana con­cept.… It’s as if we had one con­cept for the writ­ten word ‘banana’, another con­cept for pic­tures of bananas, and another con­cept for the spo­ken word ‘banana’ – but these weren’t linked in any way. The cen­tral ques­tion in today’s paper choice is this: Can we cre­ate a uni­fied deep learn­ing model to solve tasks across mul­ti­ple domains?”
  • Uber AI Labs Open Sources Pyro, a Deep Prob­a­bilis­tic Pro­gram­ming Lan­guage
    •  Pyro is a tool for deep prob­a­bilis­tic mod­el­ing, uni­fy­ing the best of mod­ern deep learn­ing and Bayesian mod­el­ing. The goal of Pyro is to accel­er­ate research and appli­ca­tions of these tech­niques, and to make them more acces­si­ble to the broader AI community.”
  • AutoML on AWS
    • In this arti­cle, we present an AWS based frame­work which allows non tech­ni­cal peo­ple to build pre­dic­tive pipelines in a mat­ter of hours while achiev­ing results that rival solu­tions hand­crafted by data scientists.”
  • https://github.com/slundberg/shap
    • Explain the out­put of any machine learn­ing model using expec­ta­tions and Shap­ley values.
  • Deep Learn­ing: AlphaGo Zero Explained In One Picture
    • Alphago Zero Cheat Sheet