Machine Learning Patents
Machine learning, deep learning, and data science are rapidly becoming key competitive differentiators. Broad patents are being issued to innovators who apply machine learning principles to specific problems.
You may believe that your machine learning solution isn't patentable because it uses many standard tools in a novel way. However, my regular review of recent machine learning patents shows that novel arrangements of standard tools can be patentable subject matter. For example, machine learning innovations have been distinguished based on the source and type of data sets used in training. In other patents, the orientation of visual data has been the distinguishing element.
Because there hasn't been as much machine learning art for examiners to draw on, machine learning patents are often very broad. Recently issued machine learning patents how that this is changing, and it will likely become harder to distinguish machine learning inventions. Still, there are always elements in even the most standard of machine learning innovations that can be patentable.
Machine learning patents focus on how a computer is trained. In contrast artificial intelligence patents protect specific algorithms.
In addition to preparing and prosecuting machine learning patents, I also enjoy hanging out with a group of our local data scientists and working on Kaggle problems, mostly in python, pandas, and tensorflow. I am not a data scientist, but I will understand your invention and can help you protect it. Below is code from one of my projects.