Test CircleCount PRO now!
Login now

Not your profile? Login and get free access to your reports and analysis.

Julien Amelot

Julien Amelot 

Science is a way of thinking much more than it is a body of knowledge. C Sagan

Occupation: Software Engineer & Data Scientist (Geico)

Location: Washington, DC

Followers: 12,529

Following: -

Views: 1,767,165

Added to CircleCount.com: 01/13/2012That's the date, where Julien Amelot has been indexed by CircleCount.com.
This hasn't to be the date where the daily check has been started.
(Update nowYou can update your stats by clicking on this link!
This can take a few seconds.
)

Tags

Sign in

The following tags have been added by users of CircleCount.com.
You can login on CircleCount to add more tags here.

  • Atheism
  • Information Technology
  • Research
  • Robotics
  • Scientists
  • Software Engineers
  • STEM

Are you missing a tag in the list of available tags? You can suggest new tags here.

Login now

Do you want to see a more detailed chart? Check your settings and define your favorite chart type.

Or click here to get the detailed chart only once.

Julien Amelot has been at 1 events

HostFollowersTitleDateGuestsLinks
Google Science Fair3,089,559The winners of Google Science Fair 2012 will be announced on July 23, 2012. This year’s 15 finalists, all 13-18 years of age, join us from seven countries, and their projects explore topics as varied as adapting hydroponic solutions for poor subsistence farmers to developing cell-phone compatible electrocardiograph machines.  This year our winners will win once-in-a-lifetime experiences like a scientific trip to the Galapagos Islands with National Geographic Expeditions, unique scholarships and real-life work opportunities in iconic science centers of excellence like CERN in Switzerland. Who will they be? Watch the awards gala live stream next Monday, July 23 at 7 p.m. PST at http://google.com/sciencefair. Google Science Fair 2012 Award Ceremony Live Stream2012-07-24 04:00:00294  

Shared Circles including Julien Amelot

Shared Circles are not available on Google+ anymore, but you can find them still here.

The Google+ Collections of Julien Amelot

108702543814354616577 has no public Google+ Collections yet.

Activity

Average numbers for the latest posts (max. 50 posts, posted within the last 4 weeks)

1
comments per post
0
reshares per post
3
+1's per post

463
characters per posting

Top posts in the last 50 posts

Most comments: 11

posted image

2014-09-15 15:34:08 (11 comments; 1 reshares; 7 +1s)Open 

Leash-free cheetah learns to jump over obstacles [2 minutes]

Most reshares: 3

posted image

2014-10-10 18:41:00 (2 comments; 3 reshares; 9 +1s)Open 

Here is Musk announcing Autopilot and Dual Motor.

#innovation   #dreamcar  

Most plusones: 17

posted image

2014-11-10 23:13:10 (1 comments; 0 reshares; 17 +1s)Open 

Ted Cruz making the headline again, this time showing his ignorance about Net Neutrality and comparing it to Obamacare ... Enjoy

#netneutrality   #theoatmeal    #crabtacos

Latest 50 posts

posted image

2016-04-28 23:52:56 (2 comments; 0 reshares; 3 +1s)Open 

UFO sighted! Or wizardry?

#RC #Helicopter

UFO sighted! Or wizardry?

#RC #Helicopter___

posted image

2016-04-15 01:55:42 (1 comments; 0 reshares; 1 +1s)Open 

Interstellar-y Robot!

Interstellar-y Robot!___

posted image

2016-04-04 01:34:43 (0 comments; 0 reshares; 5 +1s)Open 

Biggest leak ever?
11 million files, 2.7TB, investigated by 370 journalists over a year...

The website looks very well done:
https://panamapapers.icij.org/
https://panamapapers.icij.org/graphs/
https://panamapapers.icij.org/the_power_players/

A massive leak of documents exposes the offshore holdings of 12 current and former world leaders and reveals how associates of Russian President Vladimir Putin secretly shuffled as much as $2 billion through banks and shadow companies.

The leak also provides details of the hidden financial dealings of 128 more politicians and public officials around the world.

The cache of 11.5 million records shows how a global industry of law firms and big banks sells financial secrecy to politicians, fraudsters and drug traffickers as well as billionaires, celebrities and sports... more »

Biggest leak ever?
11 million files, 2.7TB, investigated by 370 journalists over a year...

The website looks very well done:
https://panamapapers.icij.org/
https://panamapapers.icij.org/graphs/
https://panamapapers.icij.org/the_power_players/

A massive leak of documents exposes the offshore holdings of 12 current and former world leaders and reveals how associates of Russian President Vladimir Putin secretly shuffled as much as $2 billion through banks and shadow companies.

The leak also provides details of the hidden financial dealings of 128 more politicians and public officials around the world.

The cache of 11.5 million records shows how a global industry of law firms and big banks sells financial secrecy to politicians, fraudsters and drug traffickers as well as billionaires, celebrities and sports stars.


___

posted image

2016-04-03 23:49:35 (0 comments; 0 reshares; 1 +1s)Open 

Successful Third flight for New Shepard (re-launch of the same booster)

New Shepard flew again on April 2, 2016 reaching an apogee of 339,178 feet or 103 kilometers. It was the third flight with the same hardware. We pushed the envelope on this flight, restarting the engine for the propulsive landing only 3,600 feet above the ground, requiring the BE-3 engine to start fast and ramp to high thrust fast.

Successful Third flight for New Shepard (re-launch of the same booster)

New Shepard flew again on April 2, 2016 reaching an apogee of 339,178 feet or 103 kilometers. It was the third flight with the same hardware. We pushed the envelope on this flight, restarting the engine for the propulsive landing only 3,600 feet above the ground, requiring the BE-3 engine to start fast and ramp to high thrust fast.___

posted image

2016-03-17 16:32:04 (0 comments; 0 reshares; 2 +1s)Open 

Robots learning Hand-Eye coordination via deep-learning

Excited to show you our toys ... and the friendly robots that help us pick them! This is a very exciting collaboration between my team and [X]. The field of computer vision has been making giant progress these past few years. As researchers, we're getting to the point where we have to look beyond pure perception, and what we can do with these newfound visual abilities. Robotics is a natural next step in that journey, linking perception and action to perform end-to-end tasks seamlessly in a robust and scalable manner. Check out the videos!___Robots learning Hand-Eye coordination via deep-learning

posted image

2016-03-17 16:23:36 (3 comments; 1 reshares; 8 +1s)Open 

The DC metro was shutdown for a whole day yesterday, with  less than a day of notice.
This visualizes the car traffic, the uber price surge and the Bikeshare status!

The DC metro was shutdown for a whole day yesterday, with  less than a day of notice.
This visualizes the car traffic, the uber price surge and the Bikeshare status!___

posted image

2016-03-17 16:16:58 (1 comments; 0 reshares; 2 +1s)Open 

Angular Material 2 is in alpha, with 6 initial components released :) #ng2

Angular Material 2 is in alpha, with 6 initial components released :) #ng2___

posted image

2016-03-15 12:48:24 (0 comments; 0 reshares; 3 +1s)Open 

AlphaGo wins the 5th game and concludes the best-of!-three by winning the tournament with an sizable lead.

Final Score AlphaGo 4  vs  1 Lee Sedol

Congrats to Lee Sedol and the DeepMind team for their performance!

AlphaGo is the first AI to beat a professional Go player. It beat the European champion last October, and just won against one of the strongest human players.
More info on AlphaGo: https://deepmind.com/alpha-go.html

Chech out the full streams or the 15 minutes recaps: https://www.youtube.com/watch?v=6hROM_bxZ9E

AlphaGo wins the 5th game and concludes the best-of!-three by winning the tournament with an sizable lead.

Final Score AlphaGo 4  vs  1 Lee Sedol

Congrats to Lee Sedol and the DeepMind team for their performance!

AlphaGo is the first AI to beat a professional Go player. It beat the European champion last October, and just won against one of the strongest human players.
More info on AlphaGo: https://deepmind.com/alpha-go.html

Chech out the full streams or the 15 minutes recaps: https://www.youtube.com/watch?v=6hROM_bxZ9E___

posted image

2016-03-09 13:57:37 (0 comments; 0 reshares; 2 +1s)Open 

Historic victory for AI yesterday: AlphaGo beat the Go world champion Sedol.
This is the 1st game of 5.
Looking forward the next games! #deepmind   #go  

AlphaGo takes the first game against Lee Sedol. In a game filled with complex fighting, they were neck and neck all the way to the end. Lee Sedol made many aggressive moves, but AlphaGo did not back down from the fights → goo.gl/O2c480___Historic victory for AI yesterday: AlphaGo beat the Go world champion Sedol.
This is the 1st game of 5.
Looking forward the next games! #deepmind   #go  

posted image

2016-03-07 19:30:38 (0 comments; 0 reshares; 1 +1s)Open 

AlphaGo vs Lee Sedol
First match will live stream: March 8th at 11pm EST
#deepmind  

Our program AlphaGo will play legendary Go player Lee Sedol in a few hours. On my way to Seoul to witness this epic event with the team !!! Now what is actually going to happen? Find out by watching live stream with english commentary here: https://www.youtube.com/watch?v=vFr3K2DORc8 First game will start on Wednesday March 9, 5 am FR time / 4 am UK time.___AlphaGo vs Lee Sedol
First match will live stream: March 8th at 11pm EST
#deepmind  

posted image

2016-02-24 15:16:21 (0 comments; 0 reshares; 0 +1s)Open 

Matei on Spark 2.0, with a planned release of May 2016.
2.0 will introduce significant performance and API improvement, with minor API breakage.
Teaser: merge of the Dataset and Dataframe APIs, introduction of structured streaming, improved code generation leading to faster data processing.

#apachespark #sparksummit  

Matei on Spark 2.0, with a planned release of May 2016.
2.0 will introduce significant performance and API improvement, with minor API breakage.
Teaser: merge of the Dataset and Dataframe APIs, introduction of structured streaming, improved code generation leading to faster data processing.

#apachespark #sparksummit  ___

posted image

2016-02-24 00:29:02 (1 comments; 0 reshares; 4 +1s)Open 

Boston Dynamics continues to improve Atlas, and to bully it :)

A new version of Atlas, designed to operate outdoors and inside buildings. It is electrically powered and hydraulically actuated. It uses sensors in its body and legs to balance and LIDAR and stereo sensors in its head to avoid obstacles, assess the terrain and help with navigation. This version of Atlas is about 5' 9" tall (about a head shorter than the DRC Atlas) and weighs 180 lbs.

Boston Dynamics continues to improve Atlas, and to bully it :)

A new version of Atlas, designed to operate outdoors and inside buildings. It is electrically powered and hydraulically actuated. It uses sensors in its body and legs to balance and LIDAR and stereo sensors in its head to avoid obstacles, assess the terrain and help with navigation. This version of Atlas is about 5' 9" tall (about a head shorter than the DRC Atlas) and weighs 180 lbs.___

posted image

2016-02-23 11:36:17 (0 comments; 0 reshares; 2 +1s)Open 

Spark on Google cloud graduates from beta
#Dataproc #Iaas

Managing #Hadoop and #Spark should be easier. And faster. And cheaper, too. That’s why we built Cloud Dataproc. Now that it’s generally available, it’s a great time to get started: http://goo.gl/2XocLP___Spark on Google cloud graduates from beta
#Dataproc #Iaas

posted image

2016-01-28 14:02:20 (0 comments; 0 reshares; 2 +1s)Open 

AlphaGo's elo ranking compared to other AIs and the European champion Fan Hui

Can a computer master the ancient game of Go? With more possible board configurations than the number of atoms in the universe, experts have long thought it to be impossible. But a system developed by researchers at Google DeepMind, called AlphaGo, has done it.

Combining a state-of-the-art Monte-Carlo tree search with two deep neural networks, AlphaGo not only has defeated the top existing Go programs at the forefront of A.I. research, but also has defeated reigning 3-time European Go champion Fan Hui. 

The next challenge will be to play the top Go player in the world over the last decade, Lee Sedol, this March in Seoul, South Korea. To learn more, visit the Google Research blog, linked below.___AlphaGo's elo ranking compared to other AIs and the European champion Fan Hui

posted image

2016-01-27 19:35:49 (0 comments; 0 reshares; 6 +1s)Open 

Google beats European Go master 5-0... and challenges world champion in March!

#ai #deepmind

Huge news from our colleagues at DeepMind. Go was long considered a key problem for AI, because it's a game for which the 'try all combinations and see which does best' is simply not possible, and human Go players rely heavily on abstract pattern recognition to guide their decisions. This is further evidence that human-level pattern recognition is no longer a 'science' problem per-se (but still a mighty engineering problem). It's time for us to focus more on exploring higher levels of cognition.___Google beats European Go master 5-0... and challenges world champion in March!

#ai #deepmind

posted image

2016-01-23 15:55:25 (0 comments; 0 reshares; 3 +1s)Open 

Successful relaunch for New Shepard

The very same New Shepard booster that flew above the Karman line and then landed vertically at its launch site last November has now flown and landed again, demonstrating reuse. This time, New Shepard reached an apogee of 333,582 feet (101.7 kilometers) before both capsule and booster gently returned to Earth for recovery and reuse.

Successful relaunch for New Shepard

The very same New Shepard booster that flew above the Karman line and then landed vertically at its launch site last November has now flown and landed again, demonstrating reuse. This time, New Shepard reached an apogee of 333,582 feet (101.7 kilometers) before both capsule and booster gently returned to Earth for recovery and reuse.___

posted image

2016-01-23 15:43:40 (1 comments; 0 reshares; 4 +1s)Open 

Snownami as of Saturday morning (Winter Storm Jonas)

Snownami as of Saturday morning (Winter Storm Jonas)___

posted image

2015-12-25 17:34:50 (0 comments; 0 reshares; 2 +1s)Open 

Merry Newton's day! aka Merry Xmas

#BostonDynamics 

Merry Newton's day! aka Merry Xmas

#BostonDynamics ___

posted image

2015-12-04 00:19:13 (2 comments; 1 reshares; 5 +1s)Open 

In the wake of the San Bernardino shooting (14 killed + 14 injured), it is time to look back at 2015 : there has been 336 days so far in 2015, and 355 mass shootings in the US, more than 1 per day on average... A mass shooting is defined as involving 4+ victims.

Washington post:
https://www.washingtonpost.com/news/wonk/wp/2015/12/02/the-san-bernardino-mass-shooting-is-the-second-today-and-the-355th-this-year/

In the wake of the San Bernardino shooting (14 killed + 14 injured), it is time to look back at 2015 : there has been 336 days so far in 2015, and 355 mass shootings in the US, more than 1 per day on average... A mass shooting is defined as involving 4+ victims.

Washington post:
https://www.washingtonpost.com/news/wonk/wp/2015/12/02/the-san-bernardino-mass-shooting-is-the-second-today-and-the-355th-this-year/___

posted image

2015-11-05 11:45:20 (0 comments; 0 reshares; 4 +1s)Open 

Smart reply coming to Inbox later this week! Let a deep neural network respond to that email for you!

Smart Reply - a deep neural network that writes email

Replying to an emails on a mobile device can be a challenge, even for short replies. What if there were a system that could automatically determine if an email was answerable with a short reply, and compose a few suitable responses that you could edit or send with just a tap? 

Smart Reply, which will be launching later this week, is built on a pair of recurrent neural networks, one used to encode the incoming email and capture the gist of what is being said and the other to predict and compose grammatically correct possible responses. 

Head over to the Google Research blog to learn more.___Smart reply coming to Inbox later this week! Let a deep neural network respond to that email for you!

posted image

2015-11-04 13:28:46 (0 comments; 0 reshares; 3 +1s)Open 

Obama is trolling the GOP following the republican candidate debate on NBC. #putin #climatechange

Obama is trolling the GOP following the republican candidate debate on NBC. #putin #climatechange___

posted image

2015-10-31 16:18:40 (2 comments; 0 reshares; 5 +1s)Open 

Fresh morning in DC for a fun 5 km race.
#raceforeverychild

http://www.novatimingsystems.com/nova/race4everychild/results.aspx?race_id=5778

Fresh morning in DC for a fun 5 km race.
#raceforeverychild

http://www.novatimingsystems.com/nova/race4everychild/results.aspx?race_id=5778___

posted image

2015-09-19 13:20:24 (0 comments; 0 reshares; 1 +1s)Open 

Trippy trip to the grocery store.  #DeepDream #neuralnetwork  

Trippy trip to the grocery store.  #DeepDream #neuralnetwork  ___

posted image

2015-06-20 09:42:28 (0 comments; 1 reshares; 2 +1s)Open 

Who said computers can't do art? #deeplearning #ANN #impressionism
Some of those are quite amazing! Check out the gallery here: https://goo.gl/photos/fFcivHZ2CDhqCkZdA

Artificial Neural Networks have spurred remarkable recent progress in image classification and speech recognition. But even though these are very useful tools based on well-known mathematical methods, we actually understand surprisingly little of why certain models work and others don’t.

Over on the Google Research blog, we take a look at some simple techniques for peeking inside these networks, yielding a qualitative sense of the level of abstraction that particular layers of neural networks have achieved in their understanding of images. This helps us visualize how neural networks are able to carry out difficult classification tasks, improve network architecture, and check what the network has learned during training. 

It also makes us wonder whether neural networks could become a tool for artists—a new way to remix visual concepts—or perhaps even shed a little light on the roots of the creative process in general.___Who said computers can't do art? #deeplearning #ANN #impressionism
Some of those are quite amazing! Check out the gallery here: https://goo.gl/photos/fFcivHZ2CDhqCkZdA

posted image

2015-06-15 18:39:47 (1 comments; 0 reshares; 5 +1s)Open 

Full house at #SparkSummit2015

Full house at #SparkSummit2015___

posted image

2015-06-08 23:47:33 (0 comments; 1 reshares; 1 +1s)Open 

Congrats to team KAIST for winning the DARPA Robotics challenge and taking home the $2m dollars price.

The challenge was the first where robots performed without being tethered and there were plenty of hard falls, soliciting groans and laughter from the crowds at the contest in Pomona, California.
The other tasks the robots were set included getting out of a car, opening a door, drilling a hole in a wall, turning a valve and crossing rubble either by clearing a path or walking over it.

More details: http://www.bbc.com/news/technology-33038914
Video of the winner: https://www.youtube.com/watch?v=q_yfvlplJe0

Congrats to team KAIST for winning the DARPA Robotics challenge and taking home the $2m dollars price.

The challenge was the first where robots performed without being tethered and there were plenty of hard falls, soliciting groans and laughter from the crowds at the contest in Pomona, California.
The other tasks the robots were set included getting out of a car, opening a door, drilling a hole in a wall, turning a valve and crossing rubble either by clearing a path or walking over it.

More details: http://www.bbc.com/news/technology-33038914
Video of the winner: https://www.youtube.com/watch?v=q_yfvlplJe0___

posted image

2015-05-16 14:47:06 (1 comments; 0 reshares; 6 +1s)Open 

Here come the self driving cars!

The +Google Self-Driving Car Project has announced the next step of their project: this summer, a few prototype vehicles will leave the test track and hit the familiar roads of Mountain View, Calif., with safety drivers aboard. Learn more on the Official Google blog, linked below.___Here come the self driving cars!

posted image

2015-04-27 12:10:51 (0 comments; 2 reshares; 3 +1s)Open 

Wow. The avalanche that hit a Mont Everest base camp, following the devastating 7.8 earthquake in Nepal.

Wow. The avalanche that hit a Mont Everest base camp, following the devastating 7.8 earthquake in Nepal.___

posted image

2015-04-13 11:04:04 (0 comments; 0 reshares; 9 +1s)Open 

Good luck to +SpaceX​ for the controlled reentry attempt later today!

Just Read the Instructions now on location in the Atlantic in advance of tomorrow's launch, targeted for 4:33pm ET. Since our last landing attempt, the drone ship has been upgraded to tolerate more powerful ocean swells, however weather at the landing site is looking significantly better this time.
 
After Dragon and Falcon 9’s second stage are on their way to orbit, the first stage will execute a controlled reentry through Earth’s atmosphere, targeting touchdown on an autonomous spaceport drone ship approximately nine minutes after launch.
 
www.spacex.com/webcast___Good luck to +SpaceX​ for the controlled reentry attempt later today!

posted image

2015-04-07 12:28:05 (0 comments; 1 reshares; 6 +1s)Open 

John Oliver interviews Edward Snowden on government surveillance. #nsa #lastweektonight

There are very few government checks on what America’s sweeping surveillance programs are capable of doing. John Oliver sits down with Edward Snowden to discuss the NSA, the balance between privacy and security, and dick-pics.

John Oliver interviews Edward Snowden on government surveillance. #nsa #lastweektonight

There are very few government checks on what America’s sweeping surveillance programs are capable of doing. John Oliver sits down with Edward Snowden to discuss the NSA, the balance between privacy and security, and dick-pics.___

posted image

2015-04-01 11:20:50 (0 comments; 0 reshares; 12 +1s)Open 

May the Force be with you. +CERN​​ confirms its existence! :P

CERN researchers confirm existence of the Force

Physicists at the European Laboratory for Particle Physics announced today that an invisible Force permeates the universe and binds the galaxy together

http://cern.ch/go/c7dB

Image shows research in the CERN Beams department

(Image credit: Maximilien Brice and Daniel Dominguez © CERN - for terms of use see http://cern.ch/copyright )___May the Force be with you. +CERN​​ confirms its existence! :P

posted image

2015-04-01 01:29:48 (0 comments; 0 reshares; 1 +1s)Open 

Great post about fooling convolution networks and linear classifiers on ImageNet

Fooling Linear Classifiers on ImageNet
http://karpathy.github.io/2015/03/30/breaking-convnets/
new blog post with a few interpretations of the fooling ConvNets papers, along with some experiments with fooling linear classifiers. I tried to structure parts of it in ways that laymen could understand because I've been seeing quite a few misconceptions surround the topic online.

(sorry to ppl seeing this twice)___Great post about fooling convolution networks and linear classifiers on ImageNet

posted image

2015-02-20 11:57:09 (1 comments; 0 reshares; 2 +1s)Open 

This week has been... Freaking cold!

Record-breaking cold in US East
http://www.bbc.co.uk/news/world-us-canada-31538987

This week has been... Freaking cold!

Record-breaking cold in US East
http://www.bbc.co.uk/news/world-us-canada-31538987___

2015-02-12 11:11:58 (0 comments; 1 reshares; 2 +1s)Open 

There are several new ImageNet results floating around that beat my 5.1% accuracy on ImageNet. Most recently an interesting paper from Google that uses "batch normalization". I wanted to make a few comments regarding "surpassing human-level accuracy":

Optimistic human performance is ~3%
I reported 5.1%, but it is interesting to try to estimate an optimistic human performance on ILSVRC by removing what I call "silly errors":

1. Note that I trained myself with 500 images, and as I documented in my blog post and our ILSVRC paper, 18 of my errors (24% - a quarter) were due to what I consider to be "class unawareness". That means that when I look at my mistake I felt that the answer was relatively evident if I had thought of that class. If I had trained longer, it's reasonable to suppose that I would have eliminated a large chunk of... more »

There are several new ImageNet results floating around that beat my 5.1% accuracy on ImageNet. Most recently an interesting paper from Google that uses "batch normalization". I wanted to make a few comments regarding "surpassing human-level accuracy":

Optimistic human performance is ~3%
I reported 5.1%, but it is interesting to try to estimate an optimistic human performance on ILSVRC by removing what I call "silly errors":

1. Note that I trained myself with 500 images, and as I documented in my blog post and our ILSVRC paper, 18 of my errors (24% - a quarter) were due to what I consider to be "class unawareness". That means that when I look at my mistake I felt that the answer was relatively evident if I had thought of that class. If I had trained longer, it's reasonable to suppose that I would have eliminated a large chunk of these, making my error ~3.9%.
2. The other issue that I call "insufficient training data" (since I was only shown 13 images / class) is also an error that falls into this category. Without this error type, the accuracy would be ~3.6%
3. The next error I'd be willing to argue I could have prevented was the fine-grained error. In the optimistic estimate, if I was willing to spend 15 minutes / terrier instead of ~5 minutes / terrier, the error would become 3.2%.

The remainder of the errors were "multiple objects" and "incorrect annotations", which I consider to be near insurmountable to some degree.

TLDR:
- 5.1% is an error rate for a human who trained for 500 images and then spent up to ~5 minutes per image.
- About ~3% is an optimistic estimate without my "silly errors".

Human ensemble experiments
This ~3% conclusion is also consistent with our "optimistic human" experiments that were based on ~250 images (and reported on in the ILSVRC paper). We had two labelers and considered an image correct if at least one of us got it. Our optimistic human performance was 2.4%, but this is a bit noisy result due to insufficient data. Moreover, we expect that an actual human ensemble would be slightly higher error than the "optimistic human", so ~3% seems relatively consistent with this interpretation.

Top5/Top1 error
As a second point, I do think we should start to look at the top-1 accuracy a bit more. I understand that there are problems with it but I do believe that there is some signal there. For example, there are only 5 snake species, so when I saw a snake image I just lazily labeled all 5 snake types and knew I got it right somewhere in top 5. In other words, the top5 error does not test differentiating between snake types. A few other categories share this property (few fishes and car types, for example).

I don't at all intend this post to somehow take away from any of the recent results: I'm very impressed with how quickly multiple groups have improved from 6.6% down to ~5% and now also below! I did not expect to see such rapid progress. 

It seems that we're now surpassing a dedicated human labeler. And imo, when we are down to 3%, we'd matching the performance of a hypothetical super-dedicated fine-grained expert human ensemble of labelers.

My blog: 
http://karpathy.github.io/2014/09/02/what-i-learned-from-competing-against-a-convnet-on-imagenet/
The ILSVRC paper that has more details on human optimistic results:
http://arxiv.org/abs/1409.0575___

posted image

2015-02-10 02:16:48 (3 comments; 0 reshares; 13 +1s)Open 

Boston Dynamics continues to improve upon previous models, Spot looks really stable, agile and quite silent compared to BigDog. A lot lighter too at 160 lbs (80 lbs diet).

Boston Dynamics continues to improve upon previous models, Spot looks really stable, agile and quite silent compared to BigDog. A lot lighter too at 160 lbs (80 lbs diet).___

posted image

2015-02-07 19:15:57 (0 comments; 1 reshares; 13 +1s)Open 

Nineteen days after they set out to achieve one of climbing's most difficult challenges, Tommy Caldwell and Kevin Jorgeson reached the summit of the 3,000-foot rock known as El Capitan in Yosemite National Park on Wednesday, marking the first free ascent of a notoriously difficult section called the Dawn Wall.

Check out the video and full article by national geographics:
http://news.nationalgeographic.com/news/2015/01/150114-climbing-yosemite-caldwell-jorgeson-capitan/

Nineteen days after they set out to achieve one of climbing's most difficult challenges, Tommy Caldwell and Kevin Jorgeson reached the summit of the 3,000-foot rock known as El Capitan in Yosemite National Park on Wednesday, marking the first free ascent of a notoriously difficult section called the Dawn Wall.

Check out the video and full article by national geographics:
http://news.nationalgeographic.com/news/2015/01/150114-climbing-yosemite-caldwell-jorgeson-capitan/___

posted image

2015-02-07 19:14:46 (0 comments; 1 reshares; 3 +1s)Open 

Tommy Caldwell and Kevin Jorgeson are the first to free-climb the 3,000-foot Dawn Wall on El Capitan, in Yosemite.
This is a nice interactive visualisation of the route

The route up the mostly smooth granite face is widely considered to be the most difficult free-climb in the world.

Tommy Caldwell and Kevin Jorgeson are the first to free-climb the 3,000-foot Dawn Wall on El Capitan, in Yosemite.
This is a nice interactive visualisation of the route

The route up the mostly smooth granite face is widely considered to be the most difficult free-climb in the world.___

posted image

2015-01-21 02:14:50 (1 comments; 0 reshares; 0 +1s)Open 

The 2015 State of the Union is starting now.
Watch Obama deliver his 6th  #SOTU  

The 2015 State of the Union is starting now.
Watch Obama deliver his 6th  #SOTU  ___

posted image

2015-01-07 22:05:28 (0 comments; 2 reshares; 6 +1s)Open 

#JeSuisCharlie  
#CharlieHebdo  

#JeSuisCharlie  
#CharlieHebdo  ___

posted image

2015-01-07 21:43:26 (0 comments; 2 reshares; 12 +1s)Open 

protests that are taking place in Paris #CharlieHebdo  #JeSuisCharlie
7.1.2015.

protests that are taking place in Paris #CharlieHebdo  #JeSuisCharlie
7.1.2015.___

posted image

2014-11-12 19:30:04 (1 comments; 0 reshares; 9 +1s)Open 

Successful comet landing! Go ESA!

ESA’s Rosetta mission has soft-landed its Philae probe on a comet, the first time in history that such an extraordinary feat has been achieved.

http://www.esa.int/Our_Activities/Space_Science/Rosetta/Touchdown!_Rosetta_s_Philae_probe_lands_on_comet

#CometLanding   #Philae   #Rosetta   #67P  ___Successful comet landing! Go ESA!

posted image

2014-11-10 23:13:10 (1 comments; 0 reshares; 17 +1s)Open 

Ted Cruz making the headline again, this time showing his ignorance about Net Neutrality and comparing it to Obamacare ... Enjoy

#netneutrality   #theoatmeal    #crabtacos

Ted Cruz making the headline again, this time showing his ignorance about Net Neutrality and comparing it to Obamacare ... Enjoy

#netneutrality   #theoatmeal    #crabtacos___

posted image

2014-11-06 22:34:23 (0 comments; 2 reshares; 9 +1s)Open 

The latest tech gadget is coming from Amazon: Echo.
Powered by a service similar to Google Now, it is a voice controlled assistant for the family to enjoy.

#AmazonEcho  #googlenow  

The latest tech gadget is coming from Amazon: Echo.
Powered by a service similar to Google Now, it is a voice controlled assistant for the family to enjoy.

#AmazonEcho  #googlenow  ___

posted image

2014-10-29 02:30:01 (0 comments; 1 reshares; 3 +1s)Open 

Google is working on a cancer detection device.
The team, led by Andrew Conrad, is hoping to use magnetic nano-particles to detect subtle changes in biochemistry and diagnose cancer early.
#GoogleX

Google is working on a cancer detection device.
The team, led by Andrew Conrad, is hoping to use magnetic nano-particles to detect subtle changes in biochemistry and diagnose cancer early.
#GoogleX___

posted image

2014-10-10 18:41:00 (2 comments; 3 reshares; 9 +1s)Open 

Here is Musk announcing Autopilot and Dual Motor.

#innovation   #dreamcar  

Here is Musk announcing Autopilot and Dual Motor.

#innovation   #dreamcar  ___

posted image

2014-10-09 14:17:47 (0 comments; 0 reshares; 2 +1s)Open 

Google Now VS Siri VS Cortana
Each platform was asked 3086 questions.
>> Google Now wins hands down.
Full details in the article linked.

#knowledgegraph  

Google Now VS Siri VS Cortana
Each platform was asked 3086 questions.
>> Google Now wins hands down.
Full details in the article linked.

#knowledgegraph  ___

posted image

2014-10-02 20:21:53 (0 comments; 2 reshares; 7 +1s)Open 

CGI at its finest!
Impecfections make him perfect

CGI at its finest!
Impecfections make him perfect___

posted image

2014-09-15 15:34:08 (11 comments; 1 reshares; 7 +1s)Open 

Leash-free cheetah learns to jump over obstacles [2 minutes]

Leash-free cheetah learns to jump over obstacles [2 minutes]___

posted image

2014-09-06 15:34:38 (3 comments; 3 reshares; 8 +1s)Open 

In the category Extreme Experiences  [2 minutes]

In the category Extreme Experiences  [2 minutes]___

posted image

2014-09-05 18:21:17 (0 comments; 0 reshares; 4 +1s)Open 

Building a deeper understanding of images

A few weeks ago we shared the results of the 2014 ImageNet Large Scale Visual Recognition Challenge, in which a team of Googlers, GoogLeNet, used a deep learning training system to automatically classify images and detect objects from Internet sources (http://goo.gl/mrCLCH).

Today on the Google Research Blog, GoogLeNet team member +Christian Szegedy offers a deeper look at the techniques the team used to place first in the classification and detection tasks of the competition, including the use of the DistBelief infrastructure (http://goo.gl/28ufU5), which makes it possible to train neural networks in a distributed manner.

Head over to the Research blog, linked below, to learn more about the techniques that will enable even better image understanding any place where it is useful to understand what is in an image as well asw... more »

Building a deeper understanding of images

A few weeks ago we shared the results of the 2014 ImageNet Large Scale Visual Recognition Challenge, in which a team of Googlers, GoogLeNet, used a deep learning training system to automatically classify images and detect objects from Internet sources (http://goo.gl/mrCLCH).

Today on the Google Research Blog, GoogLeNet team member +Christian Szegedy offers a deeper look at the techniques the team used to place first in the classification and detection tasks of the competition, including the use of the DistBelief infrastructure (http://goo.gl/28ufU5), which makes it possible to train neural networks in a distributed manner.

Head over to the Research blog, linked below, to learn more about the techniques that will enable even better image understanding any place where it is useful to understand what is in an image as well as where things are.___

Buttons

A special service of CircleCount.com is the following button.

The button shows the number of followers you have directly on a small button. You can add this button to your website, like the +1-Button of Google or the Like-Button of Facebook.






You can add this button directly in your website. For more information about the CircleCount Buttons and the description how to add them to another page click here.

Julien AmelotFacebookLinkedInCircloscope