By Michael D. Stein, BakerHostetler (December 21, 2016, 11:09 AM EST) -- Earlier this year, I wrote an article discussing software patents and the issues created by the use of functional claim limitations.[1] I explained how such functional limitations can sometimes be unclear or overbroad, and how issues of clarity and overbreadth can, and with increasing frequency do, lead to invalidation under 35 USC §§ 101 or 112. I proposed a solution whereby structural claim limitations could be used to avoid the alleged problems created by the use of functional limitations. In the present article, I discuss the use of structural claim limitations in patents directed to inventions in the rapidly growing field of machine learning (ML). Inventions in this field should be eligible for patent protection, but for patent-eligibility purposes such inventions must be described and claimed correctly and in view of the current state of the law. In the discussion below, I explain how various types of structural limitations can be employed to strengthen patents for ML inventions. This approach will benefit applicants by producing valid and enforceable patents. It will also benefit the public by producing patents of higher quality (pursuant to the U.S. Patent and Trademark Office's Enhanced Patent Quality Initiative) and greater credibility in the industry....