Shounak Mitra, MathWorks’ Product Manager for Deep Learning Toolbox, will be presenting “everything but the training” at ODSC on Thursday, May 2nd at 2 PM in Room 202. Here are some of the highlights of the talk and why you should attend.
In AI and deep learning workflows, a lot of time is spent discussing model design and model training, but the truth is when building an AI-driven system, you’ll most likely spend most of your time on auxiliary tasks such as:
- Data Preparation
- Data Preprocessing
- Model Deployment
These tasks are time-consuming, yet essential parts of the deep learning workflow, and they will determine whether you are successful at your deep learning task – or not.
In this talk, we will discuss how MATLAB can help with these tasks, and how using MATLAB enhances the TensorFlow experience. The results of each step can quickly be shared with TensorFlow through the Open Neural Network Exchange Format (ONNX). We’ll also show how to call Python from MATLAB and vice-versa.
Data Preparation: Do you have lots of data for your deep learning model, but no labels? You’ll need to label all your data before training a successful model. Shounak will show how MATLAB labeling apps can quickly label 100s of images or signals in no time with labeling automation.
Data Preprocessing: You remember the old saying: Garbage in Garbage Out? This is the golden rule of data preprocessing before training a deep learning model. We have domain-specific capabilities which provide all the preprocessing you need for image data (image normalization, contrast enhancement, and resizing) and signal data (spectrogram and wavelet scalogram) to extract good information from your data.
Yes, there will be cat pictures!
Model Deployment: MATLAB not only allows you to bring in Models from TensorFlow, but then you can automatically generate highly optimized C/C++ and CUDA code and deploy to hardware, an interesting alternative to manually translating your algorithms into low-level code. You can also deploy the entire integrated application including pre-processing and camera integration.
Not to mention, MATLAB code generation is Fast. With GPU Coder, MATLAB is faster than TensorFlow, MXNet and PyTorch.
Bring your questions about AI, Deep learning, MATLAB, and Python, and we’ll leave time at the end for Q&A.
Editor’s note: Be sure to check out Shounak’s talk at ODSC East, “Everything but the Training: 3 Ways to Enhance TensorFlow with MATLAB.”
About the writer: Johanna Pingel, Deep Learning Product Marketing
Johanna Pingel joined the MathWorks team in 2013, specializing in Image Processing and Computer Vision applications, with a current focus on deep learning applications. She owns the MATLAB Deep Learning blog and strives to create engaging and educational content for deep learning beginners and experts. She has an M.S. degree from Rensselaer Polytechnic Institute and a B.A. degree from Carnegie Mellon University.
About the presenter: Shounak Mitra, Deep Learning Product Manager
Shounak Mitra is the Product Manager for Deep Learning at The MathWorks Inc. He has over 7 years’ experience in leading teams and developing products rooted in Artificial Intelligence, Computer Vision, Natural Language Processing, and Statistical Modeling across industry and academia. He holds two Master’s degree from the University of New Hampshire – one in Mathematics and Statistics and the other in Structural Engineering with a research focus on applying machine learning principles for vision and vibration analysis. Currently, he focuses on core command line algorithms for deep learning, interfacing MATLAB with 3P frameworks like TensorFlow and PyTorch, building apps and tools for networks designs, etc.
