Here is an interview with Charlie Parker, head of large scale online algorithms at http://bigml.com
Ajay- Describe your own personal background in scientific computing, and how you came to be involved with machine learning, cloud computing and BigML.com
Charlie- I am a machine learning Ph.D. from Oregon State University. Francisco Martin (our founder and CEO), Adam Ashenfelter (the lead developer on the tree algorithm), and myself were all studying machine learning at OSU around the same time. We all went our separate ways after that.
Francisco started Strands and turned it into a 100+ million dollar company building recommender systems. Adam worked for CleverSet, a probabilistic modeling company that was eventually sold to Cisco, I believe. I worked for several years in the research labs at Eastman Kodak on data mining, text analysis, and computer vision.
When Francisco left Strands to start BigML, he brought in Justin Donaldson who is a brilliant visualization guy from Indiana, and an ex-Googler named Jose Ortega who is responsible for most of our data infrastructure. They pulled in Adam and I a few months later. We also have Poul Petersen, a former Strands employee, who manages our herd of servers. He is a wizard and makes everyone else’s life much easier.
Ajay- You use clojure for the back end of BigML.com .Are there any other languages and packages you are considering? What makes clojure such a good fit for cloud computing ?
Charlie- Clojure is a great language because it offers you all of the benefits of Java (extensive libraries, cross-platform compatibility, easy integration with things like Hadoop, etc.) but has the syntactical elegance of a functional language. This makes our code base small and easy to read as well as powerful.
We’ve had occasional issues with speed, but that just means writing the occasional function or library in Java. As we build towards processing data at the Terabyte level, we’re hoping to create a framework that is language-agnostic to some extent. So if we have some great machine learning code in C, for example, we’ll use Clojure to tie everything together, but the code that does the heavy lifting will still be in C. For the API and Web layers, we use Python and Django, and Justin is a huge fan of HaXe for our visualizations.
Ajay- Current support is for Decision Trees. When can we see SVM, K Means Clustering and Logit Regression?
Charlie- Right now we’re focused on perfecting our infrastructure and giving you new ways to put data in the system, but expect to see more algorithms appearing in the next few months. We want to make sure they are as beautiful and easy to use as the trees are. Without giving too much away, the first new thing we will probably introduce is an ensemble method of some sort (such as Boosting or Bagging). Clustering is a little further away but we’ll get there soon!
Ajay- How can we use the BigML.com API using R and Python.
Charlie- We have a public github repo for the language bindings. https://github.com/bigmlcom/io Right now, there there are only bash scripts but that should change very soon. The python bindings should be there in a matter of days, and the R bindings in probably a week or two. Clojure and Java bindings should follow shortly after that. We’ll have a blog post about it each time we release a new language binding. http://blog.bigml.com/
Ajay- How can we predict large numbers of observations using a Model that has been built and pruned (model scoring)?
Charlie- We are in the process of refactoring our backend right now for better support for batch prediction and model evaluation. This is something that is probably only a few weeks away. Keep your eye on our blog for updates!
Ajay- How can we export models built in BigML.com for scoring data locally.
Charlie- This is as simple as a call to our API. https://bigml.com/developers/models The call gives you a JSON object representing the tree that is roughly equivalent to a PMML-style representation.
You can read about Charlie Parker at http://www.linkedin.com/pub/charles-parker/11/85b/4b5 and the rest of the BigML team at
Here is an interview with Hjálmar Gíslason, CEO of Datamarket.com . DataMarket is an active marketplace for structured data and statistics. Through powerful search and visual data exploration, DataMarket connects data seekers with data providers.
HG- DataMarket is my fourth tech start-up since at age 20 in 1996. The previous ones have been in gaming, mobile and web search. I come from a technical background but have been moving more and more to the business side over the years. I can still prototype, but I hope there isn’t a single line of my code in production!
Funny you should ask about the 10 things that have surprised me the most on this journey, as I gave a presentation – literally yesterday – titled: “9 things nobody told me about the start-up business”
* Do NOT generalize – especially not to begin with
* Prioritize – and ﬁnd a work-ﬂow that works for you
* Meet people – face to face
* You are a sales person – whether you like it or not
* Technology is not a product – it’s the entire experience
* Sell the current version – no matter how amazing the next one is
* Learn from mistakes – preferably others’
* Pick the right people – good people is not enough
* Tell a good story – but don’t make them up
I obviously elaborate on each of these points in the talk, but the points illustrate roughly some of the things I believe I’ve learned … so far 😉
Both Amazon and Google have entered the public datasets space. Infochimps has 14,000+ public datasets. The US has http://www.data.gov/
So clearly the space is both competitive and yet the demand for public data repositories is clearly under served still.
How does DataMarket intend to address this market in a unique way to differentiate itself from others.
HG- DataMarket is about delivering business data to decision makers. We help data seekers find the data they need for planning and informed decision making, and data publishers reaching this audience. DataMarket.com is the meeting point, where data seekers can come to find the best available data, and data publishers can make their data available whether for free or for a fee. We’ve populated the site with a wealth of data from public sources such as the UN, Eurostat, World Bank, IMF and others, but there is also premium data that is only available to those that subscribe to and pay for the access. For example we resell the entire data offering from the EIU (Economist Intelligence Unit) (link: http://datamarket.com/data/list/?q=provider:eiu)
DataMarket.com allows all this data to be searched, visualized, compared and downloaded in a single place in a standard, unified manner.
We see many of these efforts not as competition, but as valuable potential sources of data for our offering, while others may be competing with parts of our proposition, such as easy access to the public data sets.
Ajay- What are your views on data confidentiality and access to data owned by Governments funded by tax payer money.
HG- My views are very simple: Any data that is gathered or created for taxpayers’ money should be open and free of charge unless higher priorities such as privacy or national security indicate otherwise.
Reflecting that, any data that is originally open and free of charge is still open and free of charge on DataMarket.com, just easier to find and work with.
HG- The scene is quite vibrant, given the small community. Good teams with promising concepts have been able to get the funding they need to get started and test their footing internationally. When the rapid growth phase is reached outside funding may still be needed.
There are positive and negative things about any location. Among the good things about Iceland from the stand point of a technology start-up are highly skilled tech people and a relatively simple corporate environment. Among the bad things are a tiny local market, lack of skills in international sales and marketing and capital controls that were put in place after the crash of the Icelandic economy in 2008.
I’ve jokingly said that if a company is hot in the eyes of VCs it would get funding even if it was located in the jungles of Congo, while if they’re only lukewarm towards you, they will be looking for any excuse not to invest. Location can certainly be one of them, and in that case being close to the investor communities – physically – can be very important.
We’re opening up our sales and marketing offices in Boston as we speak. Not to be close to investors though, but to be close to our market and current customers.
Ajay- Describe your hobbies when you are not founding amazing tech startups.
HG- Most of my time is spent working – which happens to by my number one hobby.
It is still important to step away from it all every now and then to see things in perspective and come back with a clear mind.
I *love* traveling to exotic places. Me and my wife have done quite a lot of traveling in Africa and S-America: safari, scuba diving, skiing, enjoying nature. When at home I try to do some sports activities 3-4 times a week at least, and – recently – play with my now 8 month old son as much as I can.
Hjálmar Gíslason, Founder and CEO: Hjalmar is a successful entrepreneur, founder of three startups in the gaming, mobile and web sectors since 1996. Prior to launching DataMarket, Hjalmar worked on new media and business development for companies in the Skipti Group (owners of Iceland Telecom) after their acquisition of his search startup – Spurl. Hjalmar offers a mix of business, strategy and technical expertise. DataMarket is based largely on his vision of the need for a global exchange for structured data.
To know more, have a quick look at http://datamarket.com/
One of the cornerstones of the technology revolution, Stanford now offers some courses for free via distance learning. One of the more exciting courses is of course- machine learning
About The Course
This course provides a broad introduction to machine learning, datamining, and statistical pattern recognition. Topics include: (i) Supervised learning (parametric/non-parametric algorithms, support vector machines, kernels, neural networks). (ii) Unsupervised learning (clustering, dimensionality reduction, recommender systems, deep learning). (iii) Best practices in machine learning (bias/variance theory; innovation process in machine learning and AI). The course will also draw from numerous case studies and applications, so that you’ll also learn how to apply learning algorithms to building smart robots (perception, control), text understanding (web search, anti-spam), computer vision, medical informatics, audio, database mining, and other areas.
Professor Andrew Ng is Director of the Stanford Artificial Intelligence Lab, the main AI research organization at Stanford, with 20 professors and about 150 students/post docs. At Stanford, he teaches Machine Learning, which with a typical enrollment of 350 Stanford students, is among the most popular classes on campus. His research is primarily on machine learning, artificial intelligence, and robotics, and most universities doing robotics research now do so using a software platform (ROS) from his group.
- When does the class start?The class will start in January 2012 and will last approximately ten weeks.
- What is the format of the class?The class will consist of lecture videos, which are broken into small chunks, usually between eight and twelve minutes each. Some of these may contain integrated quiz questions. There will also be standalone quizzes that are not part of video lectures, and programming assignments.
- Will the text of the lectures be available?We hope to transcribe the lectures into text to make them more accessible for those not fluent in English. Stay tuned.
- Do I need to watch the lectures live?No. You can watch the lectures at your leisure.
- Can online students ask questions and/or contact the professor?Yes, but not directly There is a Q&A forum in which students rank questions and answers, so that the most important questions and the best answers bubble to the top. Teaching staff will monitor these forums, so that important questions not answered by other students can be addressed.
- Will other Stanford resources be available to online students?No.
- How much programming background is needed for the course?The course includes programming assignments and some programming background will be helpful.
- Do I need to buy a textbook for the course?No.
- How much does it cost to take the course?Nothing: it’s free!
- Will I get university credit for taking this course?No.Interested in learning machine learning-
Well here is the website to enroll http://jan2012.ml-class.org/
I am going to make a case for whether to buy or not buy Zynga, and waiting to buy Facebook instead. Of course if Mark Pincus offers you a deep discount, and Mark Zuckenberg totally goes over the top with his P/E multiple, all bets would be re-valuated.
In the interest of your time, and my personal happiness, I am going to use a fairly standard way to measure attractiveness of both these companies- notably the Porter’s Five Forces Model. I will also review the recent experiences of Groupon and LinkedIn valuation to underscore what subtle differences in culture, and reputation of founders can affect the eventual value creation or destruction in an IPO.
(to be continued)
Here is an interview with Dan Steinberg, Founder and President of Salford Systems (http://www.salford-systems.com/ )
Ajay- Describe your journey from academia to technology entrepreneurship. What are the key milestones or turning points that you remember.
Dan- When I was in graduate school studying econometrics at Harvard, a number of distinguished professors at Harvard (and MIT) were actively involved in substantial real world activities. Professors that I interacted with, or studied with, or whose software I used became involved in the creation of such companies as Sun Microsystems, Data Resources, Inc. or were heavily involved in business consulting through their own companies or other influential consultants. Some not involved in private sector consulting took on substantial roles in government such as membership on the President’s Council of Economic Advisors. The atmosphere was one that encouraged free movement between academia and the private sector so the idea of forming a consulting and software company was quite natural and did not seem in any way inconsistent with being devoted to the advancement of science.
Ajay- What are the latest products by Salford Systems? Any future product plans or modification to work on Big Data analytics, mobile computing and cloud computing.
Dan- Our central set of data mining technologies are CART, MARS, TreeNet, RandomForests, and PRIM, and we have always maintained feature rich logistic regression and linear regression modules. In our latest release scheduled for January 2012 we will be including a new data mining approach to linear and logistic regression allowing for the rapid processing of massive numbers of predictors (e.g., one million columns), with powerful predictor selection and coefficient shrinkage. The new methods allow not only classic techniques such as ridge and lasso regression, but also sub-lasso model sizes. Clear tradeoff diagrams between model complexity (number of predictors) and predictive accuracy allow the modeler to select an ideal balance suitable for their requirements.
The new version of our data mining suite, Salford Predictive Modeler (SPM), also includes two important extensions to the boosted tree technology at the heart of TreeNet. The first, Importance Sampled learning Ensembles (ISLE), is used for the compression of TreeNet tree ensembles. Starting with, say, a 1,000 tree ensemble, the ISLE compression might well reduce this down to 200 reweighted trees. Such compression will be valuable when models need to be executed in real time. The compression rate is always under the modeler’s control, meaning that if a deployed model may only contain, say, 30 trees, then the compression will deliver an optimal 30-tree weighted ensemble. Needless to say, compression of tree ensembles should be expected to be lossy and how much accuracy is lost when extreme compression is desired will vary from case to case. Prior to ISLE, practitioners have simply truncated the ensemble to the maximum allowable size. The new methodology will substantially outperform truncation.
The second major advance is RULEFIT, a rule extraction engine that starts with a TreeNet model and decomposes it into the most interesting and predictive rules. RULEFIT is also a tree ensemble post-processor and offers the possibility of improving on the original TreeNet predictive performance. One can think of the rule extraction as an alternative way to explain and interpret an otherwise complex multi-tree model. The rules extracted are similar conceptually to the terminal nodes of a CART tree but the various rules will not refer to mutually exclusive regions of the data.
Ajay- You have led teams that have won multiple data mining competitions. What are some of your favorite techniques or approaches to a data mining problem.
Dan- We only enter competitions involving problems for which our technology is suitable, generally, classification and regression. In these areas, we are partial to TreeNet because it is such a capable and robust learning machine. However, we always find great value in analyzing many aspects of a data set with CART, especially when we require a compact and easy to understand story about the data. CART is exceptionally well suited to the discovery of errors in data, often revealing errors created by the competition organizers themselves. More than once, our reports of data problems have been responsible for the competition organizer’s decision to issue a corrected version of the data and we have been the only group to discover the problem.
In general, tackling a data mining competition is no different than tackling any analytical challenge. You must start with a solid conceptual grasp of the problem and the actual objectives, and the nature and limitations of the data. Following that comes feature extraction, the selection of a modeling strategy (or strategies), and then extensive experimentation to learn what works best.
Ajay- I know you have created your own software. But are there other software that you use or liked to use?
Dan- For analytics we frequently test open source software to make sure that our tools will in fact deliver the superior performance we advertise. In general, if a problem clearly requires technology other than that offered by Salford, we advise clients to seek other consultants expert in that other technology.
Ajay- Your software is installed at 3500 sites including 400 universities as per http://www.salford-systems.com/company/aboutus/index.html What is the key to managing and keeping so many customers happy?
Dan- First, we have taken great pains to make our software reliable and we make every effort to avoid problems related to bugs. Our testing procedures are extensive and we have experts dedicated to stress-testing software . Second, our interface is designed to be natural, intuitive, and easy to use, so the challenges to the new user are minimized. Also, clear documentation, help files, and training videos round out how we allow the user to look after themselves. Should a client need to contact us we try to achieve 24-hour turn around on tech support issues and monitor all tech support activity to ensure timeliness, accuracy, and helpfulness of our responses. WebEx/GotoMeeting and other internet based contact permit real time interaction.
Ajay- What do you do to relax and unwind?
Dan- I am in the gym almost every day combining weight and cardio training. No matter how tired I am before the workout I always come out energized so locating a good gym during my extensive travels is a must. I am also actively learning Portuguese so I look to watch a Brazilian TV show or Portuguese dubbed movie when I have time; I almost never watch any form of video unless it is available in Portuguese.
Dan Steinberg, President and Founder of Salford Systems, is a well-respected member of the statistics and econometrics communities. In 1992, he developed the first PC-based implementation of the original CART procedure, working in concert with Leo Breiman, Richard Olshen, Charles Stone and Jerome Friedman. In addition, he has provided consulting services on a number of biomedical and market research projects, which have sparked further innovations in the CART program and methodology.
Dr. Steinberg received his Ph.D. in Economics from Harvard University, and has given full day presentations on data mining for the American Marketing Association, the Direct Marketing Association and the American Statistical Association. After earning a PhD in Econometrics at Harvard Steinberg began his professional career as a Member of the Technical Staff at Bell Labs, Murray Hill, and then as Assistant Professor of Economics at the University of California, San Diego. A book he co-authored on Classification and Regression Trees was awarded the 1999 Nikkei Quality Control Literature Prize in Japan for excellence in statistical literature promoting the improvement of industrial quality control and management.
His consulting experience at Salford Systems has included complex modeling projects for major banks worldwide, including Citibank, Chase, American Express, Credit Suisse, and has included projects in Europe, Australia, New Zealand, Malaysia, Korea, Japan and Brazil. Steinberg led the teams that won first place awards in the KDDCup 2000, and the 2002 Duke/TeraData Churn modeling competition, and the teams that won awards in the PAKDD competitions of 2006 and 2007. He has published papers in economics, econometrics, computer science journals, and contributes actively to the ongoing research and development at Salford.
A particularly prominent technology blogger ( see http://www.readwriteweb.com/archives/michael_arrington_the_kingmaker_who_would_be_king.php )has now formalized his status as an investor (which he did even before) while relinquishing his editorial duties (which were not much given the blog’s acquisition by AOL and its own formidable line of writers, each one of whom is quite influential). Without going into either sermon mode (thou shall not have conflict of interests) or adulatory mode (wow he sold the blog for 30 mill and now he gets another 20 mill for his funds)- I shall try and present the case for ethics and ethical lapses while as a writer.