Introduction to Apache Mahout
3 likes2,179 views
Mahout is an Apache Software Foundation project that creates scalable machine learning libraries. It addresses limitations of other open source machine learning libraries such as lack of community, documentation, scalability, or licensing. Mahout began in 2008 as a Lucene subproject and became a top-level Apache project in 2010. It makes machine learning algorithms scalable by implementing them to run on Apache Hadoop for processing massive datasets. Common algorithms included are recommender systems, clustering, and classification, which see real-world use in applications such as spam filtering, product recommendations, and photo tagging.
Introduction to Apache Mahout
- 2. Machine Learning , a branch of AI, is about construction and study of system that can learn from existing data. It is used in field like: Information retrieval Identify key topics in large collections of text Biology Linear Algebra etc.
- 3. An Apache Software Foundation project to create scalable machine learning libraries under the Apache Software License. WHY MAHOUT ? Many Open Source Machine Learning libraries either: Lack Community Lack Documentation and Examples Lack Scalability Lack the Apache License Or are not research-oriented
- 4. Began life at 2008 as sub project of Apache Lucene (search, text mining- API). Lucene commiter felt it to include as separate project and mahout absorbed Taste collaborative filtering project. At April 2010, Mahout became top level apache project
- 5. Google News sees about 3.5 million new news articles per day and clustered with other articles in minutes to deliver timely. Other eg. Picasa. Mahout makes use of hadoop. Some algorithms won’t scale to massive machine clusters but map-reduce framework like apache hadoop do. Mahout convert algorithm to work at scale on top of Hadoop.
- 6. Recommender engines (Collaborative Filtering) Clustering Classification
- 7. Extensive framework for collaborative filtering. Recommenders: -- User Based -- Item Based Online and Offline support -- Offline can utilize hadoop Used by Amazon , Facebook etc.
- 9. Clustering techniques attempt to group a large number of things together into clusters that share some similarity. K-means , Fuzzy K-means Summly app also summarize similar stories from different news site and gives a brief news on that app.(concept of Google news)
- 10. Classification techniques decide how much a thing is or isn’t part of some type or category, or how much it does or doesn’t have some attribute. Example: -- Yahoo Mail spam checker -- Facebook face detection
- 11. Mahout is young ,open source , scalable machine learning library from apache Its technique are no longer theory instead deployed to solve in real world like e- commerce, video , picture etc. Scalability being the major issue Hadoop is on rescue.