machine_learning_project

Algorithms

• Deep learning
• Ensemble
• Neural networks
• Regression
• Decision Tree
• Bayesian
• Regularization
• Rule system
• Dimension Reduction
• Instanced based
• Clustering

Deep Learning

Neural network architecture

• DNN
• CNN
• RNN
  • LSTM, GRU, Bidirectional LSTM
• EA

AI

• AI search algorithms : Dijktra search, A* search
• AI game and Rule-based system

Frameworks

• Tensorflow
• Keras
• Theano
• Neon
• Pytorch
• Caffe
• MXnet
• Microsoft Cognitive Toolkit
• DeepLearning4J

  Cloud based platforms for DL

• AWS , Azure, GCP, NVIDIA GPU Cloud
• AMI : Ec2 - These AMIs come pre-installed with deep learning frameworks, such as TensorFlow, Gluon, and Apache MXNet, that are optimized for the NVIDIA Volta V100 GPUs within Amazon EC2 P3 instances
  • AML : model building feebatch prediction, Real time prediction
• Google cloud ML Engine

Big data ML

Big Data Machine Learning
General Big Data Framework:
Big data cluster deployments frameworks
HortonWorks Data Platform (HDP)
Cloudera CDH
Amazon Elastic MapReduce (EMR)
Microsoft HDInsight
Data acquisition:
Publish-subscribe framework
Source-sink framework
SQL framework
Message queueing framework
Custom framework
Data storage:
Hadoop Distributed File System (HDFS)
NoSQL
Data processing and preparation:
Hive and Hive Query Language (HQL)
Spark SQL
Amazon Redshift
Real-time stream processing
Machine Learning
Visualization and analysis
Batch Big Data Machine Learning
H2O:
H2O architecture
Machine learning in H2O
Tools and usage
Case study
Business problems
Machine Learning mapping
Data collection
Data sampling and transformation
Experiments, results, and analysis
Spark MLlib:
Spark architecture
Machine Learning in MLlib
Tools and usage
Experiments, results, and analysis
Real-time Big Data Machine Learning
Scalable Advanced Massive Online Analysis (SAMOA):
SAMOA architecture
Machine Learning algorithms
Tools and usage
Experiments, results, and analysis
The future of Machine Learning

Production pipeline - Big data

• Cluster deployment framework : HDP, Cloudera , Amazon Elastic MapReduce,Microsoft Azure HDInsight

• Data acquisition :

  • Publish-subscribe frameworks, Source-sink frameworks,
• Datastorage
  • HDFS, NoSQL
• Data pocessing & preparation
  • Data cleansing: Involves everything from correcting errors, type matching, normalization of elements, and so on, on the raw data.
    Data scraping and curating: Converting data elements and normalizing the data from one structure to another.
    Data transformation: Many analytical algorithms need features that are aggregates built on raw or historical data. Transforming and computing those extra features are done in this step
  • Hive HSQ, SparkSQL, Amazon Redshift MPP, Real-time stream processing

Big data ML platform

• H2O ARCHITECTURE
  • fork-join
  • MapReduce
• @jamal.robinson/introduction-to-h2o-ai-1ba51a884f02"">https://medium.com/@jamal.robinson/introduction-to-h2o-ai-1ba51a884f02
  
  

Funding by AI category

• ML apps
• NLP
• Computer Vision
• Smart robot
• Virtual personal assistant
• Gesture control
• Speech recognition
• Recommendation engine
• Video content recognition
• Context aware computing
• Speech to speech translation

Tuning methods for DL networks

• Back propagation
• Learning rate decay
• Max pooling
• Long short term memory
• Continuous bag of words
• Transfer learning
• Skipgram
• Batch normalization
• Dropout
• Stochastic gradient descent

AI engine with DL libs

• Theano
• Tensorflow
• CNTK
• CaffeDL4L
• Torch
• SparkML Lib : fast and engine for large scale distributed data processing
• Apache MXNet : state of the art model CNN and LSTM - Scalable
• keras :

Public datasets

• Image : Open Images V4 Google, Microsoft , UC Berkeley
• Video : Youtube
• Text : Squad, Yelp
• Satellite data : Landsat data
• Audio : Google Audio Set, Librispeech