Skip to content

Data Science & Machine Learning Newsletter # 91

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Google releases dataset of 50M vec­tor draw­ings, open sources Sketch-RNN imple­men­ta­tion Blog Post Paper Github Repo Github Repo Google Offers Cloud-Based TPU Ser­vice for Train­ing and Deploy­ing Deep Learn­ing Mod­els We’re excited to announce that our second-generation Ten­sor Pro­cess­ing Units (TPUs) are com­ing to Google Cloud to accel­er­ate a wide range of machine learn­ing Devamını Oku […]

Data Science & Machine Learning Newsletter # 90

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Sen­ti­ment analy­sis on Twit­ter using word2vec and keras word2vec is a group of Deep Learn­ing mod­els devel­oped by Google with the aim of cap­tur­ing the con­text of words while at the same time propos­ing a very effi­cient way of pre­pro­cess­ing raw text data. This post dis­c­cusses to do sen­ti­ment analy­ses with word2vec. Fully Con­vo­lu­tional Instance-aware Seman­tic Seg­men­ta­tion FCIS Devamını Oku […]

Data Science & Machine Learning Newsletter # 89

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Machine Learn­ing over­tak­ing Big Data? Big Data surged in pop­u­lar­ity both as a field and as a pop­u­lar buzz­word around 2011.   How­ever, lately is has been los­ing its lus­ter and Gart­ner in 2015 removed Big Data from its Hype Curve, replac­ing it with Machine Learn­ing. Top 10 Machine Learn­ing Videos on YouTube, updated The top machine learn­ing videos on YouTube include lec­ture Devamını Oku […]

Data Science & Machine Learning Newsletter # 88

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group How to Use Dropout with LSTM Net­works for Time Series Fore­cast­ing Long Short-Term Mem­ory (LSTM) mod­els are a type of recur­rent neural net­work capa­ble of learn­ing sequences of obser­va­tions.  This may make them a net­work well suited to time series fore­cast­ing.  An issue with LSTMs is that they can eas­ily over­fit train­ing data, reduc­ing their pre­dic­tive skill.  Dropout is a reg­u­lar­iza­tion Devamını Oku […]

Data Science & Machine Learning Newsletter # 87

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Image-to-image trans­la­tion in PyTorch (e.g. horse2zebra, edges2cats, and more) This is an ongo­ing PyTorch imple­men­ta­tion for both unpaired and paired image-to-image trans­la­tion. Baidu’s AI just achieved Zero Shot Learn­ing As babies humans often “just learn” things, it’s a gift, but now AI’s are catch­ing us up and start­ing to learn in the same way we do and it’ll Devamını Oku […]

Data Science & Machine Learning Newsletter # 86

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group 10 tools and plat­forms for data prepa­ra­tion Data prepa­ra­tion has tra­di­tion­ally been a very man­ual task and con­sumed the bulk of most data project’s time.  … For­tu­nately the lat­est gen­er­a­tion of tools, typ­i­cally pow­ered by NoSQL tech­nolo­gies take a lot of this pain away. Deep Patient: An Unsu­per­vised Rep­re­sen­ta­tion to Pre­dict the Future of Patients from the Elec­tronic Health Devamını Oku […]

Data Science & Machine Learning Newsletter # 85

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Ideas on inter­pret­ing machine learn­ing … Do I under­stand the model and answers my machine learn­ing algo­rithm is giv­ing me? And do I trust these answers? Unfor­tu­nately, the com­plex­ity that bestows the extra­or­di­nary pre­dic­tive abil­i­ties on machine learn­ing algo­rithms also makes the answers the algo­rithms pro­duce hard to under­stand, and maybe even hard to trust. … Free Machine Devamını Oku […]

Data Science & Machine Learning Newsletter # 84

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Recur­rent Neural Net­works — A Short Ten­sor­Flow Tuto­r­ial Deep Photo Style Trans­fer Code and data for paper “Deep Photo Style Trans­fer”: https://arxiv.org/abs/1703.07511 Here are some results from the algo­rithm (from left to right are input, style and our out­put):                   Evo­lu­tion Strate­gies: Devamını Oku […]

Data Science & Machine Learning Newsletter # 83

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group Why is machine learn­ing ‘hard’? There have been tremen­dous advances made in mak­ing machine learn­ing more acces­si­ble over the past few years. … How­ever, machine learn­ing remains a rel­a­tively ‘hard’ prob­lem. There is no doubt the sci­ence of advanc­ing machine learn­ing algo­rithms through research is dif­fi­cult. Deep Learn­ing for Finance: Deep Port­fo­lios We explore the use Devamını Oku […]

Data Science & Machine Learning Newsletter # 82

You want to get updates? Please join Data Sci­ence & Machine Learn­ing Newslet­ter Linked Group How Multi­n­o­mial Logis­tic Regres­sion Model Works In the pool of super­vised clas­si­fi­ca­tion algo­rithms, the logis­tic regres­sion model is the first most algo­rithm to play with. This clas­si­fi­ca­tion algo­rithm again cat­e­go­rized into dif­fer­ent cat­e­gories. These cat­e­gories purely based on the num­ber of tar­get classes.  If the logis­tic regres­sion model used for address­ing the binary clas­si­fi­ca­tion Devamını Oku […]