Dec 13, 2012

Machine Learning poster session at Stanford University

Today I attended the Stanford machine learning class (CS 229) poster session. The class was taught by Prof. Andrew Ng and has attracted 575 students, with many from companies in the local area. Total 250 projects were presented in this session.

As I walked around, I was amazed at the breadth and depth of research projects presented, many of which has immediate practical applications. I came across posters from company employees such as Walmart labs, eBay and Apple. They apply techniques learned from this class to some interesting real-world problems.

Here are some highlights:
1. Detect web traffic anomaly on an E-commerce site, and verify the anomaly with Twitter data.
 This connects mining user sentiment on Twitter (chat about this company's problem), to tracking the web performance of a company's website. The sentiment is detected by using the AFINN dataset, which contains words with scores indicating their positive and negative sentiment.

2. Recommend TV shows to facebook users based on their profile and likes (if they are willing to share their info). The students experimented with various recommender system techniques such as SVD, item-item similarity and user-user similarity. They also tried other methods such as bipartite graph, and Naive Bayes to predict user likes. 
     Other recommender systems in this poster session:
  • Beer recommendation engine
  • Best Buy Recommendation System.
 3. Design a smarter way to model stock price change, with a modified MCMC (Markov Chain Monte Carlo) algorithm. Stock prices looks highly volatile. Traditional approach comes from particle modeling in physics. Recently MCMC approach generates better results. This project is a further improvement on MCMC modeling. 

4. Separate singing voice from background music, with the PLCA (Probabilistic Latent Component Analysis) approach. It is a large matrix operation. This approach was presented in NIPS 2006 conference, and has been shown to be much more powerful than traditional approach.

There are many many more. Poster photos are listed at  Machine Learning poster session at Google+ event.

I want to end with a fun project: using an Xbox camera to learn users' hand motions and what they type. It seems there are endless things we can apply machine learning to.  


  1. Here You can Watch nice collection of New Year Messages 2016
    New year messages 2016
    New year messages


  2. Thanks for sharing such a great info it's really help full. rugby world cup live

  3. Train your customer service people to listen and connect with customers on their terms, not yours. It will make your business grow and help you retain customers. See more Academic Poster

  4. We never stop looking for qualified essay writers to hire because they're the base of our whole service. In order to pick the best out of the best, we have devised an extremely complex and challenging test for their skills and experience!

  5. awsome Install Bluestacks offline installer on your dektop. official website link but you need to click on the “Connect Android device” in order to SHAREit PC Android nice.

  6. A amazing essay is normally hushed because open as the throe through on there. Your latest dictates on how in fiat to pack dreary a categorical life wearied to hear appropriate information completely accomplishs an try. This friendly of precises wrench piece a multiple ample easier.rn positions

  7. good But we happen to be certainly not ready to carry out official website link for favorite video tutorials, a section for video tutorials SnapTube Download Free nice.

  8. The iPhone 7's exterior is similar in shape and size to iPhone 6 and iPhone 6S, though with a "mirrored" finish option. Alongside the existing silver, gold, and rose gold colors, the device is offered in new black and "jet black" colors.You can more Updates News from CydiaNerd.

  9. Buy Workshop Kits & Garage Kits Online - Storage Construction and our modular range of cupboards, wall panels, drawers and accessories that meet your requirements.