JacobiX

17 ms·

151.
by JacobiX 8y ago
Softimage|XSI : 3D animation, modeling and rendering. It was used in the film, video game and advertising industries as a tool to generate digital characters, environments and visual effects. It has a unique user interface with some pretty
152.
by JacobiX 8y ago
Page Rank is an incredibly elegant algorithm. But in practice I suppose that the actual Google search algorithm has little resemblance with the mathematically elegant PR algorithm, because, it has to implement many custom tweaking, tuning,
153.
by JacobiX 9y ago
Most modern neural networks implementations are parallelized. And that is why we can run them extremely well on the GPU. For example Volta GPUs delivers 5X increase in deep learning training compared to prior generation NVIDIA Pascal archit
154.
by JacobiX 9y ago
> hardware progress is slowing down quickly) I would be interested to know more about this. I haven't heard yet that the progress of GPUs cores for example is declining quickly ...
155.
by JacobiX 9y ago
PyTorch for example is often called a deep learning framework. But at its core it is a hardware accelerated linear algebra and tensors library with the support of auto differentiation. So it is well suited for implementing numeric algorith
156.
by JacobiX 9y ago
Face ID sensor will projects 30000 infrared dots onto the user's face, and then it reads the projected pattern. The pattern is sent to a deep neural network in order to confirm a match with the phone owner's face.
157.
by JacobiX 9y ago
Very impressive, but it has also some limitations, the softmax output that shows the confidence is often unreliable (tested empirically). This is an example of misclassification that has a score of 99%: https://imgur.com/a&#
158.
by JacobiX 9y ago
Just to confirm for the facial recognition/ detection, modern DNN algorithms outperform the 'classic' methods that took decades of continuous improvement ...
159.
by JacobiX 9y ago
Yes ONNX support is not mature even for PyTorch --> Caffe2. There are problems with basic things like affine BatchNormalization. But still the situation is a lot better now ... before we had to write our own converters from one framework
160.
by JacobiX 9y ago
It is our framework of choice especially when prototyping and implementing new differentiable programs. Although we use Caffe2 in production right now mainly because of Windows support for some of our customers. But both of them support ONN
161.
by JacobiX 9y ago
The article mention the name of a new startup too many times to be convincing.
162.
by JacobiX 9y ago
>> So the brain has a million times more the processing power and uses a million times less power. A commodity smartphone is millions of times more powerful than all of NASA's combined computing in 1970 and more energy efficient.
163.
by JacobiX 9y ago
I mostly agree that the results are not always good. As someone who read the original style transfer paper, and implemented it, I disagree that it is applying a mechanical effect. Maybe I'm biased but the algorithm use a pre-trained CN
164.
by JacobiX 9y ago
As you can see from the output, the model can simplify the shape, alter the geometry, put a texture on an object (projecting a texture like UV mapping) without being explicitly programmed to do so. It is inherently different from Photoshop
165.
by JacobiX 9y ago
Even in relatively new fields like machine learning, nearly every software that we use at work is written in C or is a wrapper around a C library, or in some cases in Fortran (a 60 years old programming language) : we use extensively PyTorc
166.
by JacobiX 9y ago
> I've come up with a method that relies on maths that most programmers would not be familiar with I'm curious about the kind of math you have used for solving the problem.
167.
by JacobiX 9y ago
Pytorch is awesome especially with automatic differentiation. We used to do the training in torch and then convert the trained model to caffe. The only problem is that the conversion script supports only torch 7. I heard that at Facebook th
168.
by JacobiX 9y ago
My bad, you are correct. I just wanted to point out that one can have concurrency without parallelism.
169.
by JacobiX 9y ago
One can have the kind of unsafe shared memory like C++ in OCaml by using a library like netmulticore. But the OCaml formalised memory model is offering the same kind of OCaml safety in multicore programs. For your first point I'm not
170.
by JacobiX 9y ago
Another great concurrency library in OCaml is the Async library. It's very convenient compared to Lwt. You can find here how it behaves with I/O operations https://realworldocaml.org/v1/en/html/concu
171.
by JacobiX 9y ago
In OCaml there is Lwt. It provides very light-weight cooperative threads. The context switches are very fast and composing cooperative threads allows the writing of highly asynchronous applications. For instance, in the versions of Go pri
172.
by JacobiX 9y ago
F# is basically an OCaml clone. I wonder why it isn't called NCaml or something like that. F# lacks some functional programming features the powerful macro system used for creating domain specific languages and extending the ocaml lang
173.
by JacobiX 9y ago
Popular recommendation algorithm like collaborative filtering by matrix factorization takes into account the accounts for user and item biases (the simplest method is to normalize the ratings of a particular user by the average of ratings o
174.
by JacobiX 9y ago
I second on this. The multicore support is not far away. Right now we have a memory model for multicore OCaml and even a multicore OCaml ARM64 backend.
175.
by JacobiX 9y ago
A lot of scientific computing is done in python (numpy) and even in Lua (using torch the great tensor library). Much of the computations are done in the GPU anyway.
176.
by JacobiX 9y ago
We were just beginning to implement some linear algebra routines in ocaml for our library that parses text into structured data. This is a very good news. We will contribute back if we have any enhancements.
177.
by JacobiX 9y ago
Locality sensitive hash is well known technique for approximate nearest search and dimensions reduction. Combining it with dnn may reduce the complexity of the dataset. But I don't know if in this setting the parameters of the lsh are
178.
by JacobiX 9y ago
Vector representation are useful for many natural language processing tasks. In word embedding like word2vec, GLoVE, fasttext for each word the algorithm learns an associated vector in dimension n (x1, x2, .., xn). A word maybe close to ano
179.
by JacobiX 9y ago
We used PNaCl to port some games to a Smart TV. It was a painful experience. Debugging chrome from a custom gdb, breaking changes from release to release, and some very subtle bugs in pepper API. But the final results were surprisingly not
180.
by JacobiX 9y ago
They can use other people's contact lists (users on LinkedIn who shared their contacts) and you surely will be on some of them.

More ›