TensorFlow is a great new deep learning framework provided by the team at Google Brain. It supports the symbolic construction of functions (similar to Theano) to perform some computation, generally a neural network based model. Unlike Theano, TensorFlow supports a number of ways to feed data into your machine learning model. The processes of getting data into a model can be rather annoying, with a lot of glue code. TensorFlow tries to fix this by providing a few ways to feed in data.

About a year ago I wrote a blog post about recent research in Deep Learning for Natural Language Processing covering several subareas. One of the areas I didn't cover was Deep Learning for Named Entity Recognition – so here are some interesting recent (2015-2016) papers related to that:

SEOUL--Humanity has fallen to artificial intelligence in checkers, chess, and, last month, Go, the complex ancient Chinese board game. But some of the world's biggest nerds are confident that machines will meet their Waterloo on the pixelated battlefields of the computer strategy game StarCraft. A key reason: Unlike machines, humans are good at lying. StarCraft, created in 1998, is one of the world's most popular computer game franchises. It pits three races against one another: the humanlike Terrans, the slimy insectoid Zerg and a mystical race with psionic powers called the Protoss.

When I studied Artificial Intelligence in the 90s, its manifestation in real life was confined to clunky machines playing chess and robots slowly navigating their way around a maze. Ideas such as singularity – where machines become capable of recurring self-improvement, their intelligence surpassing our own intelligence and all current human control or understanding – was just a concept, reality felt like several generations into the future, if at all. But the speed of AI progress has accelerated – my thesis in 1994 studied human-machine interaction via linguistics and speech recognition – this is now a standard and widely-available AI application in modern day life. My current professional interest in technology focuses on recruitment, an industry where we imagine it is very difficult to replace humans with machines due to the social and emotional interaction that is necessary at multiple levels. However AI is already shaping the way recruitment takes place, because it is starting to know more about candidates, the companies they are joining, and the work they need to do, making the use of data matching much more powerful and usefully practical than ever before.

Startup Zero Zero Robotics just took the wraps off its eye in the sky, the Hover Camera. The company hasn't set a price but expects the lightweight drone (it weighs in at 240 grams) to cost under US 600. The flying camera is a relatively new type of gadget. It all started about a year ago, when startup Lily Camera came out of stealth with its 500 to 1000 camera drone and argued that it wasn't so much a drone as a simple-to-use flying camera. This March, drone-maker DJI introduced the Phantom 4, with autonomous flying and tracking features that essentially make it that company's first flying camera at 1400.

Twitter has admitted that as many as 23 million (8.5%) of its user accounts are autonomous Twitterbots. Many are there to increase productivity, conduct research, or even have some fun. Yet many have been created with harmful intentions. In both cases, the bots have been known to behave with questionable ethics. Twitterbots, however, are minor specimens of Artificial Intelligence.

Machine Learning, we all have heard it and heard it ears full. Yet we hesitate to get a hang of it. Q) Why is it that we are scared to take a peek into it? Ans) Maybe what it does seem like a miracle to us. So we assume it is something out of our scope of learning/understanding.

To build human-like machines that can demonstrate ingenuity and creativity, the race is on to develop next generation of advanced AI (Artifical Intelligence). AI is already tackling complex tasks like stock market predictions, research synthesis etc, and'smart manufacturing' is becoming a reality where deep learning is paired with new robotics and digital manufacturing tools. Prof. Hod Lipson, director of Creative Machines Lab at Columbia University, has embarked upon exploring a higher level of AI and develop biology-inspired machines that can evolve, self-model, and self-reflect - where machines will generate new ideas, and then build them. To build self-aware robots is the ultimate goal. Prof. Lipson explains, 'Biology-inspired engineering is about learning from nature, and then using it to try to solve the hardest problems.

Artificial intelligence is getting smarter. In March, an A.I. platform named AlphaGo beat a human champion in the game of Go. Now the machines are prepping for their next big challenge: first-person shooter video games. Later this year, the 2016 Computational Intelligence and Games (CIG) Conference will host an event in which bots will pummel each other in classic "Doom," the 1993 blockbuster that established the template for a generation of action games. The CIG event will feature two "tracks": a limited match on a map known to the participants beforehand, in which bots can arm themselves with rocket launchers, and a full match on an unknown map, with every game weapon and item available.

A summary is provided in the section below. I pick some popular imagenet models, and I clock the time for a full forward backward pass. I ignored dropout and softmax layers. The CuDNN benchmarks are done using Torch bindings. One can also do the same via Caffe bindings or bindings of any other library.