October Restarts For Data Analysis, Machine Learning and Algorithms |
Written by Sue Gee | |||
Friday, 21 October 2016 | |||
If you missed an interesting MOOC the first time around, or started it and didn't finish, you are likely to get a second chance if you stay tuned. Here's a mixed bag of highly recommended classes that are now available for study. Future Learn's Learn to Code for Data Analysis opened for the second time on October 10, 2016. It is a 4-week course from the Open University aimed at non-programmers who want to be able to analyze big data sets: This hands-on course will teach you how to write your own computer programs, one line of code at a time. You’ll learn how to access open data, clean it and analyse it and to produce visualisations. You will also learn how to write up and share your analyses, privately or publicly. The computer language students get to learn is Python but rather than attempt to teach Python they are supplied with code to edit and build on in Jupyter notebooks, formerly known as iPython notebooks and widely used in research. Everything needed for the course - Python, panda (Python Data Analysis Library) module and data - has been packaged together for convenience and students are provided with two free methods of accessing it, Sage MathCloud or Anaconda. Judging from student's comments the latter option presents quite a few hurdles. On the other hand students who encounter difficulties can expect a quick response from course mentors and from other students. Although it good to keep to the timetable from the point of view of joining in discussions, which is actively encouraged, once you have joined the course you can complete it at your own pace. The course materials, including assignments are entirely free and on successful completion a certificate of participation available to purchase if you want one. As it title suggest, Machine Learning for Musicians and Artists is targeted at a specific audience. It is on Kadenze. the online arts education platform and comes from Goldsmiths, University of London. It is a 7-week course, although you can take longer over it if needed and, according to its description: In this course, students will learn fundamental machine learning techniques that can be used to make sense of human gesture, musical audio, and other real-time data. The focus will be on learning about algorithms, software tools, and best practices that can be immediately employed in creating new real-time systems in the arts. Specific topics of discussion include: • What is machine learning? • Common types of machine learning for making sense of human actions and sensor data, with a focus on classification, regression, and segmentation • The “machine learning pipeline”: understanding how signals, features, algorithms, and models fit together, and how to select and configure each part of this pipeline to get good analysis results • Off-the-shelf tools for machine learning (e.g., Wekinator, Weka, GestureFollower) • Feature extraction and analysis techniques that are well-suited for music, dance, gaming, and visual art, especially for human motion analysis and audio analysis • How to connect your machine learning tools to common digital arts tools such as Max/MSP, PD, ChucK, Processing, Unity 3D, SuperCollider, OpenFrameworks • Introduction to cheap & easy sensing technologies that can be used as inputs to machine learning systems (e.g., Kinect, computer vision, hardware sensors, gaming controllers). Some programming experience is recommended to make the most of the hands-on aspects of the course and for this Kandenze has courses introducing Processing (see Nature Of Code MOOC From Processing Foundation) and ChucK. Kadenze also offers a course on Tensor Flow - which might make a good follow on the the Future Learn Learn to Code for Data Analysis as it too uses Python/Jupyter. This is just coming to end of its first presentation with the next one starting on October 25th, 2016 and running through February 2017. It has attracted good reviews on Class Central including this one giving it 4 stars out of 5 by a student still studying it: Fun and insightful combination of learning TensorFlow with example applications using neural networks for image analysis, visualisation plus generating text and music. The course presents a minimal amount of theory, and has a hands-on approach. A typical session involves building and running a basic deep network for a task using TensorFlow commands (in Python on Jupyter notebook), getting a feel for what that does, then a guided use of a more advanced model. Assignments start with a 90% complete notebook, with gaps to fill and parameters to adjust. This one come from a student who had completed the course and awarded 5 stars: Fantastic course, moves very fast, huge learning curve, non programmers might struggle, be prepared for a lot of reading, the instructor is 1 on 1 with every student in the forums, when you finish you will realize you have entered a new universe of possibilities. You can take Kadenze courses for free and this includes access to its discussion forums which are generally well used and well run. However to participate in assignments and earn a grade and a certificate you'll need a Premium Membership subscription, which costs $10 per month or $99 for a full year. This covers full access to all Kadenze courses, of which there are over 40 with more being added. October also saw the return of Geoffrey Hinton's Neural Networks with Machine Learning from the University of Toronto, a class which I'd completed on its initial presentation and wanted to be able to recommend. An initial visit to the re-run as soon as it opened left me disappointed at it appeared to have no forum for exchange of ideas and support with any problems - something that is essential for a course of this complexity. At some point since then things have changed. Not only is there a Discussion Group it has plenty of Mentors who are responding to problems. Also it has changed its status from free Self Paced with no certificate, to a 15-week course (ending January 16) with the option of a Certificate, which there is still time earn since even though tests for the first two weeks are now marked overdue there's no penalty for late submission. There is also a strong chance the course will re-open at intervals and students will be able to carry marks forward to a new session. Another classic Coursera class had already transitioned to Coursera's new platform has now changed its format. Back in August we reported that Tim Roughgarden's highly popular, Stanford originated, two-part MOOC on the the Design and Analysis of Algorithms was up and running. This is still the case, but now it has morphed into a Specialization. Explaining this change of course (pun-intended) the instructor emailed students this information: Parts 1 and 2 of Algorithms: Design and Analysis will be re-launching as an Algorithms specialization on October 10th, 2016. The new specialization will cover the same material as these two courses, with a few additional assignments, spread out over 16 weeks (4 four-week courses) rather than 12 weeks. I hope that the somewhat slower pace and additional milestones will enable more learners to complete all of the material. Over the years, many learners have told me that there is just too much material each week to keep up with. The original version of the course was optimized for full-time undergraduates, while many learners in these online courses have jobs, families, and other responsibilities. As always, the content of the specialization will be freely available; payment is required only to receive a grade and a certificate. Unlike other Coursera specializations there is no final capstone project but if you want to follow the free route you'll need to sign up to audit the following four courses, all of which are at intermediate level.
The first of the series starts next on October 24th and being part of a Specialization it will be available on a repeating basis. It comes highly recommended. Part 1 was one of the very first offerings on Coursera. and has had well over half a million enrollments and both parts are highly rated on Class Central, where you will find many reviews. More InformationLearn to Code for Data Analysis Machine Learning for Musicians and Artists Introduction to Programming for Musicians and Digital Artists Creative Applications of Deep Learning with TensorFlow Neural Networks with Machine Learning Related ArticlesNew Learn To Code Course From Future Learn Stanford Algorithm MOOCs Relaunched Coursera Relaunches Classic Computer Science Courses Hidden Benefits of Online Machine Learning
To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.
Comments
or email your comment to: comments@i-programmer.info
|
|||
Last Updated ( Wednesday, 30 November 2016 ) |