sabalaba

Lambda Labs | Software Engineer (iOS / Computer Vision) | Palo Alto, CA (University Ave.) | Onsite | Full Time

We're hiring a Software Engineer with iOS and Computer Vision experience. Math, EE, CS, or Physics degree. If you love math and software engineering, you'll fit right in. Required experience:

- Experience shipping multiple Objective-C apps that are currently available in the app store.

- An aptitude for design, ability to innovate using all of the tools available for iOS Ideal candidates will have:

- Multiple apps in the app store.

- A track record of making high quality mobile user experiences. - Proficiency with basic computer vision and image processing operations.

To apply, please email your resume and github link to: s[at]lambdal.com

Is there research that suggests working from home is equally efficient?

Lambda Labs | Machine Learning Engineer | Palo Alto, CA (University Ave.) | Onsite | Full Time

We're hiring for the full time position of Machine Learning Research Engineer. Math, EE, CS, Physics and Neuroscience PhDs / dropouts / candidates / postdocs and curious individuals with skills in deep learning and machine learning are welcome to apply. If you love math and software engineering, you'll fit right in.

Required experience:

- Extensive experience training convolutional neural networks with gradient descent

- Strong Math and CS fundamentals: Linear Algebra, Probability, Multivariate Calculus, Mathematical optimization

Ideal candidates will have:

- Excellent academic record

- Demonstrated ability to create and implement state-of-the- art research

- GPU programming (Cuda/OpenCL/Theano)

- Programming / Mathematics Olympiads: IMO, IOI, Putnam

===

Lambda Labs | Software Engineer (iOS / Computer Vision) | Palo Alto, CA (University Ave.) | Onsite | Full Time

We're hiring a Software Engineer with iOS and Computer Vision experience. Math, EE, CS, or Physics degree. If you love math and software engineering, you'll fit right in.

Required experience:

- Experience shipping multiple Objective-C apps that are currently available in the app store.

- An aptitude for design, ability to innovate using all of the tools available for iOS

Ideal candidates will have:

- Multiple apps in the app store.

- A track record of making high quality mobile user experiences. - Proficiency with basic computer vision and image processing operations.

To apply, please email your resume and github link to: s[at]lambdal.com

If anybody is interested in playing around with this on a laptop without a GPU or without otherwise installing CUDA / torch / etc. I have a site that runs a similar algorithm on GPUs in the cloud:

https://dreamscopeapp.com/editor

The other unintended consequence of not doing any kind of material recognition is that the system will treat a plastic bag or piece of styrofoam the same as it would a large piece of sheet metal. One of those requires evasive maneuvers and the others do not.

I remember running across the NetPBM format when I was looking at some of the images that are stored in jwz's folder on the CMU AI repository. There's something very nice about storing a bitmap in pure text. (AWK image processing comes to mind.)

http://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/lang/l...

During the few years that I spent doing business in both Beijing and Shenzhen, I can say with confidence that the majority of companies there are creating innovative new technology and are not simply "copying and pasting". Also, technology startups are not an "ideas business". They are an ideas with amazing world-class execution business. The latter is what determines the bulk of the value created by the company and is much, much harder to come by.

I wonder why they're using a "magic shoebox" instead of a call market a la http://ai.eecs.umich.edu/people/wellman/?p=40.

Well, the nets aren't trained on stereoscopic data so one shouldn't expect them to have very good depth perception.

There are plenty of reasonably priced apartments outside of SF and Palo Alto. It means you might need to live in the east bay or somewhere less expensive south of the city but it's possible and maybe necessary in the early days.

Python with Theano is surprisingly expressive and performant.

To be fair, at least the DMCA gives the benefit of the doubt to content hosts via the safe harbor provisions.

It's not clear that the linked patent applies to the technique in the original post. https://www.google.com/patents/US7787664 and https://www.google.com/patents/US8712189 both seem more relevant.

Just because something looks like it could be patent encumbered doesn't mean you shouldn't pursue it. Patents need to be fought over in court. It costs millions of dollars to bring a patent lawsuit. In the event that your little side project or startup becomes a target for a patent lawsuit, frivolous or otherwise, you've likely already won.

Even with a semi-broken patent system, the world simply favors active, high-quality execution over passive "intellectual property" generation. Focus 364 days a year on a building a great product that people love and then spend a day to talk to your attorney about patents, IP, and how to protect the valuable company that you've built by focusing on your customers.

Jeff Schox has a great talk on patents and startups that is very relevant to this discussion: https://www.youtube.com/watch?v=hyv75d4JG7c

Wow, I didn't realize that this was part of a series until I saw your comment, Thanks! For everybody else who's interested:

http://norvig.com/ipython/

If you don't have ipython notebook installed, you can view them by copying the URL (without the protocol) into:

http://nbviewer.ipython.org/url/<URL>

No hypothesis is unjust! It could also be related some of the experiences people have in sensory deprivation tanks. Your brain attempting to see structure in noise and hallucinates. One hypothesis would be that on LSD, and other psychoactive substances, this feedback loop is somehow enhanced. There might be a few doctorates to be earned in testing these hypotheses.

Lambda Labs | Machine Learning Engineer | Palo Alto, CA (University Ave.) | Onsite | Full Time

We're hiring for the full time position of Machine Learning Research Engineer. Math, EE, CS, Physics and Neuroscience PhDs / dropouts / candidates / postdocs and curious individuals with skills in deep learning and machine learning are welcome to apply. If you love math and software engineering, you'll fit right in.

Required experience:

- Extensive experience training neural networks with gradient descent

- Strong Math and CS fundamentals: Linear Algebra, Probability, Multivariate Calculus, Mathematical optimization

Ideal candidates will have:

- Excellent academic record

- Demonstrated ability to create and implement state-of-the-art research

- GPU programming (Cuda/OpenCL/Theano)

- Programming / Mathematics Olympiads: IMO, IOI, Putnam

For full job descriptions and more information:

https://lambdal.com/jobs-at-lambda-labs

To apply, please email your resume and github link to:

s[at]lambdal.com

Is the federal government's inability to recruit and retain top talent really a mystery? It's surprising that an institution that is capable of confiscating double digit percentages of it's people's labor and capital gains[1] is unwilling to pay market or above market salaries to its employees.

If they want to capture the best and brightest, they'll have to do more than just appeal to patriotism.

[1] http://www.taxpolicycenter.org/taxfacts/displayafact.cfm?Doc...

Lambda Labs, located in downtown Palo Alto. We're hiring for a full-time position of Machine Learning Research Engineer. Math, EE, CS, Physics and Neuroscience PhDs / dropouts / candidates / postdocs and curious individuals interested in deep learning and machine learning are welcome to apply. If you love math and software engineering, you'll fit right in.

Required experience:

- B.S. Computer Science, Mathematics, Electrical Engineering, Physics, Neuroscience, or other quantitative field

- Extensive experience training neural networks with gradient descent

- Strong Math and CS fundamentals: Linear Algebra, Probability, Multivariate Calculus, Mathematical optimization

Ideal candidates will have:

- Excellent academic record

- Demonstrated ability to create and implement state-of-the-art research

- Published work in top journals (NIPS/ICML/ICRL)

- Experience with recurrent neural networks

- GPU programming (Cuda/OpenCL/Theano)

- Programming / Mathematics Olympiads: IMO, IOI, Putnam

For full job descriptions and more information:

https://lambdal.com/jobs-at-lambda-labs

To apply, please email your resume and github link to:

s[at]lambdal.com

To summarize, with this pre-print, Google's "FaceNet" holds the current record on performance on the LFW unconstrained face verification task with an accuracy of 0.9963 +- 0.09. Looking at the other top results on LFW, it's clear that the LFW data set is saturated.

http://vis-www.cs.umass.edu/lfw/results.html

Lambda Labs is looking for Math, EE, CS and Neuroscience PhD dropouts / candidates / post docs / or simply curious individuals interested in deep learning, machine learning, and representation learning. If you love math and software engineering, you'll fit right in.

Required experience:

- B.S. Computer Science, Mathematics, Electrical Engineering, Neuroscience, or other quantitative field.

- Training neural networks with gradient descent.

- Strong Math and CS fundamentals: Linear Algebra, Probability, Multivariate Calculus, Mathematical optimization

Desired Experience / Personality:

- Published work in top journals (NIPS/ICML/ICRL)

- Creating and implementing state-of-the-art research

- GPU programming (Cuda/OpenCL)

- Programming / Mathematics Olympiads: IMO, IOI, Putnam

To apply, please email your resume and github link to: s[at]lambdal.com

Pretty interesting, not a single mention of any of the founders appear on the S-1 form.

Lambda Labs (https://lambdal.com/jobs-at-lambda-labs) a machine learning and AI company that makes the Lambda Labs face recognition API.

Where: Palo Alto, California

You: Looking for Math, EE, CS, and Neuroscience PhD dropouts / candidates / post docs / or simply curious individuals interested in machine learning, representation learning, and deep learning with strong math fundamentals.

Desired experience:

  - Training neural networks with gradient descent.
  - Strong math fundamentals: Linear Algebra, Multivariate Calculus,
    Mathematical optimization
  - GPU programming (Cuda/OpenCL) a plus.

Interns OK!

Email your resume and github link to: s[at]lambdal.com

Here's an animation of an autoencoder learning filter weights. It's interesting that they look similar.

https://lambdal.com/images/autoencoder-learning-face-filters...

Conducting warfare with autonomous robots seems preferable to having human soldiers on the front lines. Technology like this will not only reduce costs but more importantly reduce the number of human lives lost in war.

I've had the pleasure and displeasure of working with small datasets (~7.5GB of images) in shell. One often needs to send SIGINT to the shell when it starts to glob expand or tab complete a folder with millions of files. But besides minor issues like that, command line tools get the job done.

Yea, so at Lambda, depending on the application, we use either convolutional neural networks or layered neural networks initialized with unsupervised training using autoencoders to do the feature extraction from images. The features are either passed through a softmax layer or used as a "siamese" network. (Siamese networks provide a similarity metric as opposed to a probability distribution over labels.[3])

As far as recognition goes, most modern algorithms don't use fiducials (what you call 'nodal points') to do recognition. So there's nothing in our code that has to do with interocular distance or mouth width, etc., so all of the features are "learned". However, fiducials are used to do a frontalization or re-alignment step before passing it through the feature extractor (neural network or otherwise). This type of frontalization step allowed Facebook's deepface algorithm to use locally learned filters for their CNN as opposed to shared weights.[1][2]

[1] http://www.cv-foundation.org/openaccess/content_cvpr_2014/pa...

[2] http://mmlab.ie.cuhk.edu.hk/pdf/YiSun_CVPR14.pdf

[3] http://yann.lecun.com/exdb/publis/pdf/chopra-05.pdf

Exactly, and they were suggesting that it would be $200 USD more than that per week!

Those costs in China are too high. I've lived in Shenzhen for under $70 USD / week in a hostel. If you eat out every meal for 100 RMB, you're not even going to be spending $350 a week for meals. While I agree that if you can do it locally, do it locally, but those numbers for the cost of living and travel in China are inaccurate.

Pretty alarmist writing. "I therefore can’t resist showing a new piece of Google technology: the military robot ‘WildCat’ made by Boston Dynamics which was bought by Google in December 2013". And goes on to show a video of something that was funded by DARPA well before Google purchased Boston Dynamics. Google has already said multiple times that it won't be pursuing any military contracts.

I would add "Practical recommendations for gradient-based training of deep architectures" to the list for those who already have a feel for training multi-layer neural nets. It provides a good overview for those that want to learn more about gradient descent, hyperparameter tuning, and other practical considerations involved with training deep architectures.

http://arxiv.org/abs/1206.5533