The fun thing about spending time at MIT is that you always run into interesting things. Couple of days ago, I encountered the MIT Bot submission for NASA - Sample Return Robot Challenge.
NASA and the Worcester Polytechnic Institute (WPI) in Worcester teamed up for competing in the Sample Return Robot Challenge to demonstrate a robot that can locate and retrieve geologic samples from a wide and varied terrain without human control.
Sample Return Robot Challenge is part of NASA centennial challenges; a robot which has autonomous capability to locate and retrieve specific sample types from various locations over a wide and varied terrain and return those samples to a designated zone in a reasonable amount of time with limited mapping data.
The challenge description follows:
The Sample Return Robot Challenge is scheduled for June 14-17, 2012 in Worcester, MA. The Challenge requires demonstration of an autonomous robotic system to locate and collect a set of specific sample types from a large planetary analog area and then return the samples to the starting zone. The roving area will include open rolling terrain, granular medium, soft soils, and a variety of rocks, and immovable obstacles (trees, large rocks, water hazards, etc.) A pre-cached sample and several other samples will be located in smaller sampling zones within the larger roving area. Teams will be given aerial/geological/topographic maps with appropriate orbital resolution, including the location of the starting position and a pre-cached sample.
MIT Robotics Team 2015 Promo Video
The bot is powered with the following technologies:
ROS: The Robot Operating System (ROS) is a set of software libraries and tools that help you build robot applications. From drivers to state-of-the-art algorithms, and with powerful developer tools, ROS has what you need for your next robotics project. And it's all open source. www.ros.org
Arduino: Arduino is an open-source electronics platform based on easy-to-use hardware and software. It's intended for anyone making interactive projects.
RabbitMQ for Async messaging: RabbitMQ is a messaging broker - an intermediary for messaging. It gives your applications a common platform to send and receive messages, and your messages a safe place to live until received.
MIT team couldn't make it to the challenge due to some technical issues. NASA has awarded $100,000 in prize money to the Mountaineers, a team from West Virginia University, Morgantown, for successfully completing Level 2 of the Sample Return Robot Challenge, part of the agency’s Centennial Challenges prize program.
- Robots Face Off in $1.5 Million NASA Sample Return Challenge
- NASA's robot event challenges, robots, engineers
- NASA Awards $100,000 to Winning Team of Robot Challenge
- MIT CrowdFunding Page
- MIT RoboTeam
- Non-linear classification and regression, kernels
- Passive aggressive algorithm
- Overfitting, regularization, generalization
- Content recommendation
Dr. Jaakkola's socratic method of inquiring the common sense questions ingrain the common concepts in the mind of people. The class started with the follow up of perceptron from yesterday and quickly turned into a session on when NOT to use perceptron such as in case of non linearly seperable problems. Today's lecture was derieved from 6.867 Machine Learning Lecture 8. The discussion extended to Support Vector Machine (and Statistical Learning Theory) Tutorial, which is also well explained in the An Idiot’s guide to Support vector machines (SVMs) R. Berwick, Village Idiot
Speaking of SVM and dimensionality, Dr. Jaakkola posed the question if ranking can also be a secondary classification problem? Learning to rank or machine-learned ranking (MLR) is a fascinating topic where common intuitions like number of items displayed, error functions between user's preference and display order sparseness fall flat. Microsoft research has some excellent reference papers and tutorials on learning to rank which are definitely worth pouring over in case you are interested in this topic. Label ranking by learning pairwise preferences is another topic discussed in detail during the class. Some reference papers follow:
- A Short Introduction to Learning to Rank
- Reviewing Seeing stars: exploiting class relationships for sentiment categorization with respect to rating scales
- LETOR: Learning to Rank for Information Retrieval Tutorials on Learning to Rank
- Ranking Methods in Machine Learning A Tutorial Introduction
- Yahoo! Learning to Rank Challenge Datasets
- Large Scale Learning to Rank
- Yahoo! Learning to Rank Challenge Overview
- Multiclass Classification: One-vs-all
- Zipf, Power-laws, and Pareto - a ranking tutorial Lada A. Adamic
Indeed with SVM, the natural progression led to the 'k' word; kernel functions. A brief introduction to kernel classifiers Mark Johnson Brown University is a good starting point and The difference of kernels in SVM?, and how to select a kernel for SVM provide good background material to understand the practical aspects of kernel. Kernels and the Kernel Trick Martin Hofmann Reading Club "Support Vector Machines"
The afternoon topic was Anomaly detection; use cases included aberrant behavior in financial transactions, insurance fraud, bot detection, manufacturing quality control etc. One the most comprehensive presentations on Anomaly Detection Data Mining Techniques is by Francesco Tamberi which is great for the background. Several problems worked on during the class were from 6.867 Machine learning which shows how instructors carefully catered the program for practitioners with the right contents from graduate level courses, as well as industry use cases. Other topics discussed included Linear versus nonlinear classifiers and we learned how decision boundary is the region of a problem space in which the output label of a classifier is ambiguous. Class discussions and Q&A touched on the wide variety of subjects including but not limited to How to increase accuracy of classifiers?, Recommendation Systemsm A Comparative Study of Collaborative Filtering Algorithms which eventually led to Deep Learning Tutorial: From Perceptrons to Deep Networks which performed really well on MNIST Database for handwritten digits.
- Caltech 101
- THE MNIST DATABASE of handwritten digits
- Why do naive Bayesian classifiers perform so well?
Linear vs. non linear classifiers followed where Dr. Jaakkola spoke about why logistic regression a linear classifier, more on Linear classifier, Kernel Methods for General Pattern Analysis, Kernel methods in Machine learning, How do we determine the linearity or nonlinearity of a classification problem? and review of Kernel Methods in Machine Learning
Misc. discussions of Kernel Methods, So you think you have a power law, Radial basis function kernel, Kernel Perceptron in Python surfaced, some of which briefly reviewed in Machine Learning: Perceptrons- Kernel Perceptron Learning Part-3/4. Shape Fitting with Outliers and SIGIR 2003 Tutorial Support Vector and Kernel Methods tutorial with radial basis functions. Other topics included Kernel based Anomaly Detection with Multiple Kernel Anomaly Detection (MKAD) Algorithm, Support Vector Machines: Model Selection Using Cross-Validation and Grid-Search, LIBSVM -- A Library for Support Vector Machines, Practical Guide to Support Vector Classification, Outlier Detection with Kernel Density Functions and Classification Framework for Anomaly Detection as relevant readings.
Looking forward to the Deep Learning and Boosting tomorrow! Dr. Barzilay said its going to be pretty cool.
As a follow up on MIT's tackling the challenges of Big Data, I am currently in Boston attending Machine Learning for Big Data and Text Processing Classification (and therefore blogging about it for posterity based on public domain data / papers - nothing posted here is MIT proprietary info to violate any T&C). MIT professional education courses are tailored towards professionals and it is always a great opportunity to learn what others practitioners are up to, especially in a relatively new field of data science.
Today's lecture #1 was outlined as
- machine learning primer
- features, feature vectors, linear classifiers
- On-line learning, the perceptron algorithm and
- application to sentiment analysis
Instructors Tommi Jaakkola (Bio) (Personal Webpage) and Regina Barzilay (Bio) (Personal Webpage) started the discussion with breif overview of the course. Dr. Barzilay is a great teacher who explains the concepts in amazing detail. As an early adapter and practitioner, she was one of the technology review innovator under 35.
The course notes are fairly comprehensive; following are the links to the publicly available material.
- Youtube: http://www.youtube.com/MITProfessionalEd
- FB: https://www.facebook.com/MITProfessionalEducation
- twitter: https://twitter.com/MITProfessional
- LinkedIn - https://www.linkedin.com/grp/home?gid=2352439
In collaboration with CSAIL - MIT Computer Science and AI Lab- www.csail.mit.edu, today's lecture was a firehose version of Ulman's large scale machine learning. Dr. Barzilay walked through the derivation of the Perceptron Algorithm, covering Perceptrons for Dummies and Single Layer Perceptron as Linear Classifier. For a practical implementation, Seth Juarez's NUML implementation of perceptron is a good reading. A few relevant publications can be found here.
- NLP Programming Tutorial 3 - The Perceptron Algorithm
- Machine Learning: Exercise Sheet 4
- Perceptron Find Weight
- ML LAb Solutions
- Classification Exercise
- Perceptron Learning
The discussion progressed into Opinion Mining and Sentiment Analysis with related techniques. Some of the pertinent data sets can be found here:
- Huge ngrams dataset from googlestorage.googleapis.com/books/ngrams/books/datasetsv2.html
- Global ML dataset repository: https://archive.ics.uci.edu/ml
- Sentiment 140 Dataset
- Cornell Movie Review Dataset
Dr. Barzilay briefly mentioned Online Passive-Aggressive Algorithms and details from Lillian Lee, AAAI 2008 Invited Talk - A “mitosis” encoding / min-cost cut while talking about Domain Adaptation which is quite an interesting topic on its own. Domain Adaptation with Structural Correspondence Learning by John Blitzer, Introduction to Domain Adaptation guest lecturer: Ming-Wei Chang CS 546, and Word Segmentation of Informal Arabic with Domain Adaptation are fairly interesting readings. The lecture slides are heavily inspired by Introduction to Domain Adaptation guest lecturer: Ming-Wei Chang CS 546.
With sentiment analysis and opinion mining, we went over the seminal Latest Semantic Analysis - LSI, Clustering Algorithm Based on Singular Value Decomposition, Latent Semantic Indexing (LSI), (Deerwester et al. 1990), and Latent Dirichlet Allocation (LDA), (Blei et al. 2003). The class had an interesting discussion around the The Hathaway Effect: How Anne Gives Warren Buffett a Rise, with a potential NSFW graphic. The lecture can be summed up in Comprehensive Review of Opinion Summarization Kim, Hyun Duk; Ganesan, Kavita; Sondhi, Parikshit; Zhai, ChengXiang (PDF version).
Few other papers / research work and demos discussed during the lecture included Get out the vote: Determining support or opposition from Congressional floor-debate transcripts, Multiple Aspect Ranking using the Good Grief Algorithm, Distributional Footprints of Deceptive Product Reviews, Recursive Neural Tensor Network - Deeply Moving: Deep Learning for Sentiment Analysis, Code for Deeply Moving: Deep Learning for Sentiment Analysis, and Sentiment Analysis - The Stanford NLP Demo, Stanford Sentiment Treebank.
Among several class discussions and exercises/quiz, The Distributional Footprints of Deceptive Product Reviews was of primary importance. Started with Amazon Glitch Unmasks War Of Reviewers, darts were thrown around Opinion Spam Detection: Detecting Fake Reviews and Reviewers , Fake Review Detection: Classification and Analysis of Real and Pseudo Reviews
With all this sentiment analysis talks, I have asked fellow attendee Mohammed Al-Hamdan (Data Analyst at Al-Elm Information Security Company), about publishing a paper by the end of this course on sentiment analysis in Arabic language twitter feeds for potential political dissent. Would be a cool project / publication.
Looking forward to the session tomorrow!
Bonus, here is Dr. Regina Barzilay — Information Extraction for Social Media video - publicly available on youtube.
Deep learning architectures are built using multiple levels of non-linear aggregators, for instance neural nets with many hidden layers. In this introductory talk Will Stanton discusses the motivations and principles regarding learning algorithms for deep architectures. Bill provides a primer to neural networks, and deep Learning. He explains how Deep Learning gives some of the best-ever solutions to problems in computer vision, speech recognition, and natural language processing.
and also, why Google is Investing in deep learning.
An interesting portrayal of Microservices by Martin Fowler.
The term "Microservice Architecture" has sprung up over the last few years to describe a particular way of designing software applications as suites of independently deployable services. While there is no precise definition of this architectural style, there are certain common characteristics around organization around business capability, automated deployment, intelligence in the endpoints, and decentralized control of languages and data.
Interesting introduction to Functional Programming by Kelsey Innis.
Dated, but a highly recommended reading for functional programmers by John Hughes of Institutionen Datavetenskap,
Abstract: As software becomes more and more complex, it is more and more important to structure it well. Well-structured software is easy to write, easy to debug, and provides a collection of modules that can be re-used to reduce future programming costs. Conventional languages place conceptual limits on the way problems can be modularised. Functional languages push those limits back. In this paper we show that two features of functional languages in particular, higher-order functions and lazy evaluation, can contribute greatly to modularity. As examples, we manipulate lists and trees, program several numerical algorithms, and implement the alpha-beta heuristic (an algorithm from Artificial Intelligence used in game-playing programs). Since modularity is the key to successful programming, functional languages are vitally important to the real world.
I have recently encountered the following error when enumerating through the UserPrincipal.GetAuthorizationGroups collection.
System.DirectoryServices.AccountManagement.PrincipalOperationException: An error (1301) occurred while enumerating the groups. The group's SID could not be resolved.
The problem was introduction of the domain controller running Server 2012 while the machine running my application was win7 VM (applies to Win2K8 as well)
With little googling, it appears that for the Windows 7 VM with introduction of 2012 domain controller, this SID error appears to be a known issue. When a 2012 domain controller is involved, the GetAuthorizationGroups() function would fail on groups (SIDs) that are added to a user by default.
Installing KB2830145 fixed my problem.
- GetAuthorizationGroups() Fails on Windows 2008 R2/WIN7
- StackOverflow: UserPrincipals.GetAuthorizationGroups An error (1301) occurred while enumerating the groups. After upgrading to Server 2012 Domain Controller
- KB2830145: SID S-1-18-1 and SID S-1-18-2 cannot be mapped on Windows-based computers in a domain environment