Artificial intelligence is one of the hottest areas of research right now. Large companies such as IBM, Google, Microsoft, Facebook, and Amazon have not only increased funding for their development research departments, but have also begun acquiring companies that have a strong track record in machine learning, neural network The company is a small startup in the field of natural language and image processing. Given the current explosion in the field of AI research, professors at Stanford University made this report not long ago: "The role of AI software is becoming increasingly powerful, and AI software that has a strong impact on human society and the economy will be available by 2030."

The foreign website Datamation today collated the current popular 15 open source artificial intelligence software, Lei Feng network (search "Lei Feng network" public number concern) on the full text compiled as follows.

1.Caffe

Developed by a PhD student at UC Berkeley, the Caffe software is a deep learning framework based on expression constructs with scalable coding. Popular with researchers and companies for its outstanding processing speed. According to its homepage, the software can process over 60 million images a day using just one NVIDIA K40 GPU. The Caffe software project is run by BVLC, with companies such as NVIDIA and Amazon funding its development research.

2.CNTK

CNTK stands for Computational Network Toolkit, one of Microsoft's open source artificial intelligence software. What makes it so great is that it performs very well whether only the CPU is working, a single GPU, or or multiple GPUs, or multiple machines with multiple GPUs working. Although Microsoft uses it primarily for research in speech recognition, it can also perform machine translation, image recognition, image capture, text processing, language recognition and language modeling.

3. Deeplearning4j

Deeplearning4j is a JVM open source deep learning library software. It can run in a distributed environment and integrates Hadoop with Apache Spark. And this makes it possible to configure deep neural networks and also run other JVM languages such as Java and Scala.

The Deeplearning4j software project is operated by Skymind Commercial, and Skymind is responsible for the project's financial support, staff training and product promotion.

4.DMTK

Like CNTK, DMTK stands for Distributed Machine Learning Toolkit, another open source AI software from Microsoft. Designed for big data research and designed to reduce the training time of artificial intelligence systems. DMTK consists of three main components: the DMTK framework, the LightLDA model algorithm and the distributed (multisensory) word embedding algorithm. To illustrate DMTK's processing speed, Microsoft claims that it can process a topic model containing 1 million topics and 10 million words (a total of 10 trillion parameters) in a document collection of over 100 billion tags through a cluster of 8-node computers, which is unmatched by similar software.

5. H2O

H2O, which focuses more on the enterprise experience than on scientific research, has reaped orders from large enterprises such as Capital One, Cisco, Nielsen Catalina, PayPal and Transamerica. Oxdata says everyone can use H2O's machine learning and predictive analytics capabilities to solve business problems. H2O can also be used for predictive modeling, fraud and risk analysis, insurance analytics, ad craft, healthcare and customer intelligence.

There are two open source versions of H2O: the standard version of H2O and the Sparkling Water version of H2O, both integrated in Apache Spark, which Oxdata will support for paying businesses.

6. Mahout

Mahout is an open source project under the ASF (Apache Software Foundation) that provides open source machine learning frameworks. According to its official homepage, Mahout has three main features: a scalable algorithmic programming environment, pre-written algorithms for tools such as Spark and H2O, and a built-in vector computing environment called "Samsara". Mahout user's include companies such as Adobe, Accenture, Foursquare, Intel, Collage, Twitter and Yahoo!

7. MLlib

Known for its processing speed, Apache Spark has become one of the most commonly used software for big data processing. The MLlib algorithm is a scalable machine learning library for Spark. It is integrated with Hadoop and can operate cross-platform with NumPy and NumPy.R. It encompasses a large number of machine learning algorithm types, including: classification, regression, decision trees, suggestion, aggregation, topic modeling, feature transformation, model evaluation, ML pipeline construction, ML persistence, survival analysis, frequent sets, sequential pattern mining, discrete linear algebra, and statistical analysis of data.

8. NuPIC

NuPIC, operated by Numenta, is an open source artificial intelligence project based on HTM (hierarchical temporal memory) theory. In fact, the HTM theory seeks to design a computer system based on the neocortical structure of the human brain, aiming to create a "computer that approaches or exceeds human capabilities in processing cognitive tasks. "

Numenta not only owns the open source license of NuPic, but also its commercial license and patent certificate.

9. OpenNN

OpenNNis an advanced AI software built for researchers and developers， Provide runnable neural network ofC++ Programming Library。 Its main feature is the deep framework with efficient performance。OpenNN The official website has information about neural network Detailed description of the operation。 because ofOpenNN The customer provides technical support is from SpainArtelnics company（ Major in predictive analytics research）。

10. OpenCyc

Developed by Cycorp, the OpenCyc software has a built-in Cyc knowledge base and common sense reasoning engine, including 239,000 terms, about 2,093,000 triples, and about 69,000 web ontology languages, which is like an external semantic database. It is very effective in rich domain modeling, semantic data fusion, text understanding, domain-specific expert systems, and game AI. There are two versions of OpenCyc: a version that is free and not open source for researchers, and a version that charges for businesses.

11. Oryx 2

Built on Apache Spark and Kafka, Oryx 2 is a development framework software that specializes in large scale machine learning. It uses a unique three-layer lambda construction. Developers can use Oryx 2 to develop new software with built-in procedures for a number of generic large data tasks such as collaborative filtering, classification, normalization and clustering. Big data tool provider Cloudera developed the initial Oryx 1 project and is still continuing its research and development.

12. PredictionIO

Back in February of this year, Salesforce bought the PredictionIO software program. The PredictionIO software project provides a new platform and trademark for ASF as an incubation project. While Salesforce is developing its own machine learning technology through the PredictionIO software project, it is still open source. It can use machine learning to deploy relevant web services that help users build a predictive engine by responding to dynamic requests on a page in real time.

13. SystemML

SystemML was originally developed by IBM and later acquired by ASF as its Big Data project. SystemML is a highly extensible platform for performing advanced mathematical operations and executing R or Python-like statements. SystemML runs on Spark or Hadoop and can be used for customer callbacks for 4S store vehicle maintenance, airport airspace control, and social media data collection for bank customers.

14.TensorFlow

TensorFlow is one of Google's open source artificial intelligence software, providing a library for numerical computation using data flow graphs that can run on single/multiple CPU or GPU systems and even mobile devices. Google says TensorFlow is flexible, truly portable, has automatic identification capabilities and supports Python and C++ platforms.

15.Torch

Torch describes itself as "having a scientific computing framework that supports machine learning algorithms that can prioritize the use of GPUs." Torch is characterized by its outstanding flexibility and outstanding processing speed, and is very handy for processing machine learning, computer vision, signal processing, parallel processing, images, video, audio and computer networks. It is written in LuaJIT, a scripting language based on the Lua language.

Recommended>>
1、Baidu and Changsha Intelligent Driving Research Institute cooperate to successfully build intelligent driving heavy trucks
2、Google CEO AI is more significant than fire and electricity
3、This wheelless car can go up to 550 miles per hour and fly in traffic
4、NARA partners join forces at SAICGeneral MotorsPanasia Software Quality Conference
5、Revealed How Oeanc the Oceanic Chain is opening the way to ocean trade

已发送

发送中