Subir Mansukhani

492 Flips | 1 Magazine | 71 Likes | 10 Followers | @SubirMansukhani | Keep up with Subir Mansukhani on Flipboard, a place to see the stories, photos, and updates that matter to you. Flipboard creates a personalized magazine full of everything, from world news to life’s great moments. Download Flipboard for free and search for “Subir Mansukhani”

Historic Achievement: Microsoft researchers reach human parity in conversational speech recognition - The AI Blog

Microsoft has made a major breakthrough in speech recognition, creating a technology that recognizes the words in a conversation as well as a person …

Deep Reinforcement Learning Through Policy Optimization, John Schulman, OpenAI

Slides from a recent presentation by John Schulman of OpenAI on Deep RL through Policy Optimization, a promising artificial intelligence technique.

Machine Learning

[1610.02391] Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization

Computer Science > Computer Vision and Pattern Recognition<p>Title: Grad-CAM: Visual Explanations from Deep Networks via Gradient-based …

Bayesian Correlation Estimation: Dealing with Uncertainty and Non-Stationarity

Authors: Sepideh Sadeghi and Thomas Wiecki<br>Correlation is a critical concept in assessing and minimizing risk in a portfolio. Often the goal is to …

Data Science

General

README.md<p>The Dynamic Neural Network Toolkit<p>General<p>Installation<p>C++<br>• Python<p>Getting Started<p>Citing<p>Releases and Contributing<p>DyNet is a neural network …

Crash: how computers are setting us up for disaster | Tim Harford

We increasingly let computers fly planes and carry out security checks. Driverless cars are next. But is our reliance on automation dangerously diminishing our skills?<p>When a sleepy Marc Dubois walked into the cockpit of his own aeroplane, he was confronted with a scene of confusion. The plane was …

Automation

Watch all six episodes of the series Trust Disrupted: Bitcoin and the Blockchain

Our latest video series explores the world of bitcoin and the blockchain, and you can now watch all six episodes on YouTube and TechCrunch.com.<p>The series features Nathaniel Popper, a <i>New York Times</i> reporter and the author of the book on which the series is based, along with several bitcoin …

These industrial robots teach each other new skills while we sleep

It takes days to reprogram an industrial robot. With artificial intelligence, it could take only a few hours.<p>Fanuc, maker of the industrial robots used to assemble Apple’s iPhone and cars for Volkswagen and Tesla, is now partnering with Nvidia to add the company’s graphics processing units to its …

Small Data is the New Big Data

How many times did you apologize this week? How often did you have negative thoughts about yourself? What about directed to others? Which sounds did …

Infographics

Interactive Demos

README.md<p><b>Run Keras models in the browser, with GPU support using WebGL</b><p>Interactive Demos | Documentation<p>Run Keras models in the browser, with GPU …

Deep Learning

Top 10 IPython Notebook Tutorials for Data Science and Machine Learning

<b>By Matthew Mayo, KDnuggets.</b><p>This post is made up of a collection of 10 Github repositories consisting in part, or in whole, of IPython (Jupyter) …

How to Use t-SNE Effectively

Although extremely useful for visualizing high-dimensional data, t-SNE plots can sometimes be mysterious or misleading.

Google's Deep Mind Gives AI a Memory Boost That Lets It Navigate London's Underground

The Deep Mind AI lab is giving deep learning a memory boost Photo: iStockphotoGoogle’s DeepMind artificial intelligence lab does more than just …

[1610.03995] Semi-Supervised Active Learning for Support Vector Machines: A Novel Approach that Exploits Structure Information in Data

Statistics > Machine Learning<p>Title: Semi-Supervised Active Learning for Support Vector Machines: A Novel Approach that Exploits Structure Information …

Uncertainty in Deep Learning (PhD Thesis) | Yarin Gal

<i>Function</i> draws from a dropout neural network. This new visualisation technique depicts the distribution over <i>functions</i> rather than the predictive …

Abhinav Rastogi, Flipkart - Handling performance for Progressive Web Apps at scale

Understanding empirical Bayesian hierarchical modeling (using baseball statistics)

<b>Previously in this series:</b><p>Suppose you were a scout hiring a new baseball player, and were choosing between two that have had 100 at-bats each:<p>A …

Data Science

CS 294-131 Special Topics in Deep Learning Fall 2016

Lecture 1<p>8/31/16<p>Kaiming He: <b>Deep Learning Gets Way Deeper</b> [slides, introductory slides]<p>Reading:<p>Deep Learning, by Yann LeCun, Yoshua Bengio, and …

Apparent Display Resolution Enhancement for Moving Images

Apparent Display Resolution Enhancement for Moving Images, Siggraph 2010<p><b>Apparent Display Resolution Enhancement for Moving Images</b><p>1 MPI Informatik 2 …

Adobe Acrobat

Towards deep symbolic reinforcement learning

Towards deep symbolic reinforcement learning Garnelo et al, <i>2016</i>Every now and then I read a paper that makes a really strong connection with me, one …

Machine Learning

Analyze Your Experiment with a Multilevel Logistic Regression using PyMC3​

<i>Note: In this post, I assume some familiarity with PyMC. If you need to get up to speed in a hurry and you’re familiar with linear regression, go</i> …

Allison Gilmore, Data Scientist, Ayasdi @ MLconf SF

HistWords: Word Embeddings for Historical Text

HistWords:<br>Word Embeddings for Historical Text William L. Hamilton, Jure Leskovec, Dan Jurafsky<p>Introduction<p>HistWords is a collection of tools and …

Using Machine Learning to Predict Out-Of-Sample Performance of Trading Algorithms

Remote Processing Using Futures

A new version of the future package has been released and is available on CRAN. With futures, it is easy to <i>write R code once</i>, which later <i>the user can</i> …

Model evaluation, model selection, and algorithm selection in machine learning

Almost every machine learning algorithm comes with a large number of settings that we, the machine learning researchers and practitioners, need to …

Algorithms

10/36

<b>Collin Eubanks (Head TA)</b> <b><br>Email:</b> ceubanks@andrew.cmu.edu <b><br>Office Hours:</b> Thursdays 12-1 BH 232K<p><b>Chenghui Zhou</b> <b><br>Email:</b> chenghuz@andrew.cmu.edu <b><br>Office Hours:</b> …