OpenTalks.AI - Олег Абраменко, Patents: AI & ML

-

Transcript

1. PATENTS : AI & ML Oleg Abramenko Managing partner, IT PATENT info@itpatent.ru
2. Some statistics… ▪ ~380 000 granted patents (75 000 patent families) ▪ ~20 000 granted patents in 2016 ▪ ~94 000 granted patents in 2017!!
3. Number of AI patents granted by country
4. Number of AI patents granted by technology
5. What about Russia? ▪ ~2936 granted patents in 2016 ▪ ~400 granted patents in 2017 ▪ Top Russian assignees: Yandex, ABBYY, Kaspersky Lab… ▪ Examples: – Training convolutional neural network on graphics processing units (RU2424561, Microsoft) – Intellectual controller with neuron network and self- modification rules (RU2266558, Kuban State Technological University)
6. Obtain patent: US, EPO, RU US Practice ▪ Alice Corp. v. cls Bank – patentable subject matter (patent eligibility) ▪ McRo v. Bandai – Claims with more specific limitations are more likely to pass muster because they cannot be unduly abstracted and are less likely to invoke preemption concerns. – The improvement in a computer does not need to relate to the electronic or mechanical operation of the computer itself (e.g., the efficiency of the processor), but can relate to the improvement in accuracy and effect provided by the computer. ▪ IDS (Information Disclosure Statement)
7. Obtain patent: US, EPO, RU EPO Practice ▪ Two identities/Comvik (T 641/00) – Technical but not inventive ▪ Auction method/Hitachi (T 258/03) – Technical but not inventive ▪ Circuit simulation I/Infineon Technologies (T 1227/05) – Technical and inventive
8. Obtain patent: US, EPO, RU ▪ Rospatent practice ▪ EAPO practice
9. Worry about broad patents? Google has a US940617 on a common machine learning technique called dropout. This means that Google could insist that no one else use this technique until 2032.
10. What’s going wrong? Some patents describes applying well-known technique X in domain Y. US5944839, for a “system and method for automatically maintaining a computer system.” This patent includes very broad claims applying AI to diagnosing problems with computer systems: A method of optimizing a computer system, the method comprising the steps of: detecting a problem in the computer system; activating an AI engine in response to the problem detection; utilizing, by the AI engine, selected ones of a plurality of sensors to gather information about the computer system; determining, by the AI engine, a likely solution to the problem from the gathered information; and when a likely solution cannot be determined, saving a state of the computer system.
12. Oleg Abramenko Managing partner, IT PATENT info@itpatent.ru http://itpatent.ru