alexbw

Osmo | osmo.ai/careers | Full-time | Machine learning engineer, data analyst, in NYC and other roles

We have digitized vision, and hearing, but not scent — our oldest and deepest sense. Join Osmo on our mission to give computers a sense of smell to improve human health and happiness.

Thrive | Associate Director / Director, Machine Learning | Cambridge, MA | Full-time | Onsite | https://thrivedetect.com/come-to-thrive

Thrive is passionate about our mission to integrate earlier #cancer detection to empower longer, healthier lives. We are seeking a highly motivated Associate Director/Director, Machine Learning to join our Biostatistics and Informatics team in Cambridge, MA. The successful candidate will be responsible for developing Machine Learning strategies with multi-omics and other types of data, to enable new molecular diagnostic products. The Associate Director/Director will possess deep statistical knowledge, strong analytical skills, solid scientific background, and a high level of professional leadership.

More info here: https://www.linkedin.com/jobs/view/1739089430/

Actually, advanced autodiff is one of its intended points of, er, differentiation :). The authors wrote the original Autograd package [0], released in 2014, that led to “autograd” becoming used as a generic term in PyTorch and other packages. JAX has all of the autodiff operations that Autograd does, including `grad`, `vjp`, `jvp`, etc. We’re working on the number of supported NumPy ops, which is limited right now, but it’s early days. Try it out, we’re really excited to see what you build with it!

0: https://github.com/hips/autograd

Check out github.com/google/jax, it’s NumPy on the GPU with automatic differentiation, JIT and autobatching.

+1 for more varied examples

Neat stuff. Are data structures up to the programmer, or are there primitives for e.g. matrices?

Directly addresses the "dark art" of hyperparameter tuning in machine learning.

Poked around the word cloud, but the output of the clustering doesn't seem to many any sense to me.

"boat" is by "scarecrow", "adopted", and "feelings"

"window" is by "insulted", "prize" and "arson"

Rotating around the word-in-question didn't pop up any more similar words.

t-SNE is known for producing immediately interpretable clusterings, but this seems a bit obscure.

@tuananh I've got the dataset stored away, but I don't know if I'm legally allowed to post it. Would love if someone could produce proof one way or the other.

@viraj_shah I spent about 6 months working on the project before I had to stop to concentrate on my schoolwork (I was a senior in collge at the time). I think it would have been impossible to do this for myself without Cython. If it were to happen today, I would probably be writing in PyCuda, or with Numba, and it would be much, much, MUCH more succinct.

You'd use Novocaine to grab the audio, and then you'd do the audio format conversion yourself. e.g.:

[[Novocaine audioManager] setInputBlock:^(float * inData, UInt32 numSamples, UInt32 numChannels) { for (int i=0; i < numSamples * numChannels; ++i) { dataForVoip[i] = YourType(inData[i]) * someScaling; } }];

Fixed.

Documentation will be forthcoming. What do you mean by audio sample formats? One of the things I tried to do was make sure that you never, ever, EVER have to think about anything but floating point audio. In my experience, audio comes in as either float or SInt16s (I've never seen fixed-point coming in by default, but I haven't done an enormous amount of odd-peripheral Mac audio), and novocaine just handles it for you.

Hey everybody, Novocaine is my baby, and it's completely awesome to see folks already using it.

For the record, it's open-source, I just forgot to stick the MIT license at the top. Or would folks prefer BSD? Let me know, I'm flexible. The only thing I care about is that people use it to make awesome software.

If anybody's a Boston local, I'll be talking about iOS audio (both novocaine and some fancier frequency analysis, I hope) at the Berklee College of Music CS club tomorrow, 6-8pm, at 150 Mass Ave, room 118.

And, if you dig novocaine, check out the apps I made with it: http://itunes.apple.com/us/artist/alex-wiltschko/id344345862

Sounds intriguing, just signed up. Don't have an excuse to not try to bang out some work that's been sitting on the back burner.