Proficiency in Neural Networks
Wolfram Language provides users with many tools for importing, modifying and training neural networks. Completing the Level 1 certification exercises demonstrates that you have achieved proficiency in using these tools for a range of machine learning tasks. This set of certification exercises covers cleaning and synthesizing training data, optimizing training parameters and options, implementing transfer learning and even implementing adversarial machine learning.
You'll Need to Know About...
- Network Training: Choosing a reasonable network capacity and addressing problems like overfitting via different means.
- Data Cleaning: Dealing with missing and messy data so that it can be used for network training.
- Data Encoding: Taking different types of input data and encoding them into vectors and tensors for use in a neural network.
- Network Architecture: Constructing and modifying neural networks on a layer-by-layer basis.
- Transfer Learning: Modifying existing networks and retraining them to perform new tasks.
- Adversarial Networks: Creating a network that can produce inputs that fool another network.
Certification Policies
Certification exercises are provided in a Wolfram Cloud notebook (which may also be downloaded). Candidates select two of the four exercises; write full, properly commented solutions; and submit their completed exercises for grading within one week. A score of 70% or higher is required for Level 1 certification. Candidates may request accommodations for special requirements due to disabilities by contacting Wolfram U, who will notify them of a decision. Your exercise notebook should not be shared with anyone else. Failure to follow these policies may result in the disqualification of a candidate for certification.
Recommended Level 1 Certification Process
- Familiarize yourself with the Wolfram Language neural network framework. A good place to start is the guide page, which links to many of the relevant functions as well as to related tech notes, workflow guides and more. The Wolfram Neural Networks Boot Camp provides a hands-on immersive training in the subject.
- Find additional resources in the Wolfram U catalog. Use the available filters and search to find courses specific to the topics you want to study.
- If you are unfamiliar with neural networks, you may want to start with the video lesson Neural Networks Basic Concepts. You can get a better sense of neural networks in Wolfram Language specifically by watching this gentle introduction to the neural net framework and/or by attending our Introduction to Neural Networks course.
- Pay the $195 certification fee. (Education customers pay $100; students pay $50.) This certification fee is included with the boot camp full registration fee.
- Download the notebook and choose two of the four possible exercises to solve. (Only two will be graded, so don't do more.) Fill out the notebook with your solutions. Make sure to explain your code and confirm that it runs properly in a fresh kernel.
- Submit your notebook to us for grading within one week using the file upload link that is provided when you pay your fee.
- Your Level 1 certificate will be awarded once we grade your notebook and assign it a passing score.