EEDSP
21.9K views | +2 today
Follow
EEDSP
Digital Signal Processing, Data Analytics, Big Data, HPC, Deep Learning, GPGPU, Distributed and Parallel Computing, Natural Language Processing
Curated by Shiwon Cho
Your new post is loading...
Your new post is loading...
Scooped by Shiwon Cho
Scoop.it!

Announcing Tensor Comprehensions –

Announcing Tensor Comprehensions – | EEDSP | Scoop.it
Today, Facebook AI Research (FAIR) is announcing the release of Tensor Comprehensions, a C++ library and mathematical language that helps bridge the gap between researchers, who communicate in terms of mathematical operations, and engineers focusing on the practical needs of running large-scale models on various hardware backends. The main differentiating feature of Tensor Comprehensions is…
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

ParlAI

ParlAI | EEDSP | Scoop.it
ParlAI (pronounced “par-lay”) is a framework for dialog AI research, implemented in Python.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Caffe 2

NVIDIA and Facebook announced Caffe2. 

  

Cafe2: FA New Lightweight, Modular, and Scalable Deep Learning Framework

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Introducing DeepText: Facebook's text understanding engine

Introducing DeepText: Facebook's text understanding engine | EEDSP | Scoop.it
DeepText can understand with near-human accuracy the textual content of several thousand posts per second, spanning more than 20 languages.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Dragon: A distributed graph query engine

Dragon: A distributed graph query engine | EEDSP | Scoop.it

Dragon, a distributed graph query engine, is the next step in our evolution toward efficient query plans for more complex queries. 

Dragon monitors real-time updates to the social graph modeled as objects and associations and creates several different types of indices that improve the efficiency of fetching, filtering, and reordering the data. With Dragon, far less data is transferred over the network to the web tier, improving latency and freeing up CPU for other computation.Dragon monitors real-time updates to the social graph and creates several different types of indices that improve the efficiency of fetching, filtering, and reordering the data.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Fighting spam with Haskell

Fighting spam with Haskell | EEDSP | Scoop.it
We recently completed a two-year redesign of Sigma, one of our spam-fighting systems. Check out how we integrated Haskell with our existing C++ code and the improvements we made to GHC.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Open-sourcing Facebook Infer: Identify bugs before you ship

Open-sourcing Facebook Infer: Identify bugs before you ship | EEDSP | Scoop.it
Facebook Infer is a static program analyzer that Facebook uses to identify bugs before mobile code is shipped.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

FAIR open sources deep-learning modules for Torch

FAIR open sources deep-learning modules for Torch | EEDSP | Scoop.it
The modules are significantly faster than the default ones in Torch and have accelerated research projects by allowing users to train larger neural nets in less time.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook HydraBase adds reliability to Hadoop's HBase

Facebook HydraBase adds reliability to Hadoop's HBase | EEDSP | Scoop.it
Facebook revamps its HBase implementation for faster recovery from downtime and higher reliability
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

WebScaleSQL | "We're Gonna Need A Bigger Database"

WebScaleSQL | "We're Gonna Need A Bigger Database" | EEDSP | Scoop.it

WebScaleSQL is a collaboration among engineers from several companies that face similar challenges in running MySQL at scale, and seek greater performance from a database technology tailored for their needs.

Our goal in launching WebScaleSQL is to enable the scale-oriented members of the MySQL community to work more closely together in order to prioritize the aspects that are most important to us. We aim to create a more integrated system of knowledge-sharing to help companies leverage the great features already found in MySQL 5.6, while building and adding more features that are specific to deployments in large scale environments. In the last few months, engineers from all four companies have contributed code and provided feedback to each other to develop a new, more unified, and more collaborative branch of MySQL.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Under the Hood: Building and open-sourcing RocksDB

Under the Hood: Building and open-sourcing RocksDB | EEDSP | Scoop.it

Every time one of the 1.2 billion people who use Facebook visits the site, they see a completely unique, dynamically generated home page. There are several different applications powering this experience--and others across the site--that require global, real-time data fetching.   

 

Storing and accessing hundreds of petabytes of data is a huge challenge, and we're constantly improving and overhauling our tools to make this as fast and efficient as possible. Today, we are open-sourcing RocksDB, an embeddable, persistent key-value store for fast storage that we built and use here at Facebook.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook’s graph-processing engine

This is a good presentation about Facebook’s graph-processing engine, Giraph, from a big data event held at the company’s Menlo Park campus in early June.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook rolls out new web and database server designs • The Register

Facebook rolls out new web and database server designs • The Register | EEDSP | Scoop.it

Facebook has released under the Open Compute umbrella the design of a new database server they’ve introduced in one of the datacenters. The bit that caught my eyes is that this is not about more disk space or more CPU, but redundant power supplies:

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

facebookresearch/Detectron

facebookresearch/Detectron | EEDSP | Scoop.it
Detectron - FAIR's research platform for object detection research, implementing popular algorithms like Mask R-CNN and RetinaNet.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook Publishes New Neural Machine Translation Algorithm

Facebook Publishes New Neural Machine Translation Algorithm | EEDSP | Scoop.it
Facebook’s Artificial Intelligence Research team published research results using a new approach for neural machine translation (NMT). Their algorithm scores higher than any other system on three established machine translation tasks.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Smaller and faster data compression with Zstandard

Smaller and faster data compression with Zstandard | EEDSP | Scoop.it
With a performance-first design optimized for modern CPUs, Facebook's new compression algorithm translates directly to faster data transfer and smaller storage requirements.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Improving Facebook's performance on Android with FlatBuffers

Improving Facebook's performance on Android with FlatBuffers | EEDSP | Scoop.it
In last six months, we have transitioned most of Facebook on Android to use FlatBuffers as the storage format.
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook open sources tools for bigger, faster deep learning models

Facebook open sources tools for bigger, faster deep learning models | EEDSP | Scoop.it
Facebook on Friday open sourced a handful of software libraries that it claims will help users build bigger, faster deep learning models than existing tools allow. The libraries, which is calling modules, are alternatives for the default ones in a popular machine learning development environment called Torch, and are optimized to run on graphics processing…
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Under the Hood: Indexing and ranking in Graph Search

Under the Hood: Indexing and ranking in Graph Search | EEDSP | Scoop.it

Search ranking is the process of choosing the best set of results for a search query. Our team combines data-driven techniques with intuition to create a process that is roughly:

Come up with an idea for a ranking change – ideas come from a mix of creativity on the part of our ranking engineers and feedback from our users

Implement the ranking change, test it, and launch it to a very small fraction of our user base

Measure the impact of the ranking change

Our goal is to maximize searcher happiness, which we do our best to quantify through metrics like click through rate (CTR), NDCG, engagement, etc. We have to measure the impact of ranking changes on all of these metrics to maintain a proper balance.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook Open Sources Animation Engine Pop That Powers Paper

Facebook Open Sources Animation Engine Pop That Powers Paper | EEDSP | Scoop.it
Facebook today open sourced Pop, the engine behind Paper's smooth animations and transitions, over on GitHub. Pop uses dynamic instead of traditional stati
more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Hadoop's 10 in Facebook's 10 - hadoopsphere.com

Hadoop's 10 in Facebook's 10 - hadoopsphere.com | EEDSP | Scoop.it

As the social media giant Facebook celebrated it's 10th anniversary, let's take a look at how the company has been impacting Hadoop ecosystem. Listed below are 10 ecosystem projects in which Facebook has done significant open source contributions. Also, towards the end, listed below are some of the other Hadoop ecosystem projects which are not yet open source but occupy a position of prominence inside the company's technical environment.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Presto | Distributed SQL Query Engine for Big Data

Presto | Distributed SQL Query Engine for Big Data | EEDSP | Scoop.it

Distributed SQL Query Engine for Big Data.

Presto is an open source distributed SQL query engine for running interactive analytic queries against data sources of all sizes ranging from gigabytes to petabytes.

more...
No comment yet.
Scooped by Shiwon Cho
Scoop.it!

Facebook kisses DRAM goodbye, builds memcached for flash

Facebook kisses DRAM goodbye, builds memcached for flash | EEDSP | Scoop.it
Facebook has developed a new data cache called McDipper that’s essentially memcached rewritten to run on flash memory instead of DRAM, thus saving money while still delivering higher performance than disk.
more...
No comment yet.