Tweets
- Tweets
- Tweets & replies
- Photos & videos
@benhamner is blocked
Are you sure you want to view these Tweets? Viewing Tweets won't unblock @benhamner.
Hash instability in Julia composite types: I just hit this issue, anyone know why it's present? http://stackoverflow.com/questions/27661255/hash-instability-in-julia-composite-types …
2 retweets 1 favoriteThe more you work with real data, the more you look at statistics assumptions and think "well, that's cute"
76 retweets 78 favoritesFacebook has the richest data on human interaction that's ever existed. There should be a million scientists creating insights on it, not 20
9 retweets 15 favoritesUnderstanding the group dynamics and success of teams: analyzing online collaboration in 150,000 @github projects http://arxiv.org/pdf/1407.2893v1.pdf …
Interview with @ylecun on convolutional neural nets and CIFAR-10 http://blog.kaggle.com/2014/12/22/convolutional-nets-and-cifar-10-an-interview-with-yan-lecun/ …
"When experts are wrong, it's often because they're experts on an earlier version of the world" http://paulgraham.com/ecw.html
4 retweets 6 favoritesDistinguishing cause from effect using observational data: methods and benchmarks http://arxiv.org/pdf/1412.3773v1.pdf …
18 retweets 39 favoritesDeepSpeech: @AndrewYNg's & Baidu's deep learning system for end-to-end speech translation, using RNNs trained on GPUs http://arxiv.org/pdf/1412.5567v1.pdf …
The next million startups: Machine Intelligence for X. X=industry, use case, challenge, job, service, tool, pain point, etc.
7 retweets 9 favoritesCan you use a car trip's telematic signature to identify the driver? Win $30k in the latest @kaggle competition https://www.kaggle.com/c/axa-driver-telematics-analysis …
Only 11% of business leaders think today's graduates have the requisite skills, vs. 96% of chief academic officers https://hbr.org/2014/10/the-real-revolution-in-online-education-isnt-moocs …
3 favorites"In recent cleanup of one important ML system at Google, we ripped out 10,000s of lines of unused experimental code" http://static.googleusercontent.com/media/research.google.com/en/us/pubs/archive/43146.pdf …
9 retweets 18 favoritesI wish every scientific paper included a section titled "Unsuccessful Experiments"
25 retweets 45 favoritesGreat overview of NIPS 2014 from @johnplattml: deep learning, large scale learning, and ML in production systems http://blogs.technet.com/b/machinelearning/archive/2014/12/16/machine-learning-trends-from-nips-2014.aspx …
How to lead and mislead with charts: Russia's foreign currency reserves in 2014 http://imgur.com/a/rmaN8
1 retweet 3 favoritesClassify 30k plankton images into 121 classes to monitor ocean health! $175k in prizes from @BoozAllen on @kaggle https://www.kaggle.com/c/datasciencebowl …
Distilling the knowledge in a neural network using soft targets http://www.dlworkshop.org/65.pdf
1 retweet 6 favoritesThe tragedy of the software commons http://pchiusano.github.io/2014-12-08/failed-software-economics …
1 retweet 4 favoritesHalf the papers that appeared at NIPS would have been rejected if the peer review process was rerun http://mrtz.org/blog/the-nips-experiment/ …
8 retweets 6 favoritesProduct Manager at @SmartRecruiters. Good at pattern recognition, haiku and picking chanterelles. And I think my spaceship knows which way to go.
Current working on a game called Voxel Quest. Very interested in 3D algorithms, AI, procedural generation, puppies, etc.
Twitter may be over capacity or experiencing a momentary hiccup.
Visit Twitter Status for more information.