Last December 10th, I reported on a “land rush” for “deep learning” patents. Now that we’re half way through 2020, it’s time for an update. Before I show the final result from 2019 and indicate how things are going during 2020, I’ll begin by describing my USPTO search terms.
First, I used the USPTO “advanced” engine. Second, I’ve memorized only two acronyms: ACLM is the acronym for the Claim(s) field and ISD is the acronym for the Issue Date.
The Issue Date portion is simple. Last year, it was “ISD/2019.” This year it’s “ISD/2020.”
Now for the ACLM search. It’s more complicated but has the same format: ACLM/(“deep learning” or “deep neural” or “multi-layer neural”).
The “terms” of the search are surrounded by quotation marks because each term is two words. For example, had I used ACLM/(deep learning), etc. the system would interpret the search to mean ACLM /(deep AND learning) or (deep AND neural) or (multi-layer AND neural), and the results would include patents where, in the Claims, the words deep and learning would be present but unrelated to each other.
Now let’s combine the two. The search becomes ACLM/(“deep learning” or “deep neural” or “multi-layer neural”) AND ISD/2020. At the end of last year, 2019, there were 396 patents. This year, at June 30, 2020, the number was 301.
Before I show the Bar Chart, I should note that I also went back in time, to January 1, 2001. The only blips are 2002 (4 patents), 2013 (3), 2014 (4) and 2015 (4). The other years prior to 2013 were either 0 or 1 and didn’t register on the Chart. Because there were so few patents prior to 2013, I opened each patent, scanned the Claims text, and satisfied myself that the relevant term then was “multi-layer neural.”
In the bar chart below, I use “deep learning” to include the other variations in the ACLM/ search.
So 2019 ended with 396 “deep learning” patents, and, at the half-way mark in 2020, the number is already at 301. The number of 2020 patents is likely to exceed 600. If the 2020 pace quickens, 2020 may double 2019. The year-over-year doubling since 2016 is approximately accurate.
Last year, I pointed out that, by keeping all of the patents in numerical order and building a spreadsheet from the bottom up, my search terms led organically to a leader board that put the “usual suspects” at the top: IBM, Google and Microsoft.
As of last December 3, I mentioned only the top 12. At this half-way point in 2020, I’ll be more inclusive. Here’s the leader board as of June 30, 2020.
| “Deep Learning” PATENT LEADER BOARD | # |
| 1. IBM | 74 |
| 2. Google | 50 |
| 3. Microsoft Technology Licensing, LLC | 41 |
| 4. Siemens Corporation (including Healthcare) GmbH | 39 |
| 5. Baidu | 25 |
| 6. Samsung Electronics Co. Ltd. (excludes two MOS patents, now expired) | 24 |
| 7. Adobe Systems | 22 |
| 8. Intel | 18 |
| 9. General Electric Company | 17 |
| 10. Amazon.com, Amazon Technologies, and Amazon subsidiary A9.com | 16 |
| 10.Facebook | 16 |
| 12. Ford Global Technologies LLC | 14 |
| 12. NEC | 14 |
| 14. Capital One Services, LLC | 10 |
| 14. KLA-Tencor | 10 |
| 14. Qualcomm, Inc. | 10 |
| 17. Electronics and Technology Research Institute (Elec. TRI) | 9 |
| 18. ALIBABA Group Holding | 8 |
| 18. Apple, Inc. | 8 |
| 18. Gyrfalcon Technology, Inc. | 8 |
| 18. INTRASPEXION | 8 |
| 18. Nuance | 8 |
| 18. Wipro Limited (IN) | 8 |
Nelson E. (Nick) Brestoff holds a B.S. degree in Engineering Systems from UCLA, an M.S. degree in Environmental Engineering Science from the California Institute of Technology, and a J.D. degree from the USC Gould School of Law. He was a California litigator for 38 years. He is the sole or lead co-inventor on eight deep learning patents, now assigned to Intraspexion LLC, and is author of AI Concepts for Business Applications.
Sign up for the free insideAI News newsletter.
