際際滷

際際滷Share a Scribd company logo

Don't Manage Litigation,Prevent it! Use Deep Learning

Download as PPTX, PDF
Evolve Law
Evolve Law

This document discusses how deep learning and artificial intelligence can be used to analyze large amounts of legal and company data to help reduce litigation costs. It proposes training a deep learning model on public legal records to identify risks and early warnings of potential legal issues. This could help companies examine their own internal data to surface problematic patterns and issues, and take steps to avoid litigation by addressing risks proactively. The approach aims to help companies shift away from reactive litigation management toward preventing legal problems before they arise.

1 of 16
Download to read offline
EVOLVE LAW
DARWIN
DARWIN
LITIGATION-$160 BILLION PER YEAR (2001- 2010) EVEN WHEN A BUSINESS WINS, IT LOSES ABOUT $350,000 AND $400,000 PER CASE - SO THE ONLY WAY TO WIN IS U.S. Commercial Tort Costs 0. 45. 90. 135. 180. 225. 2001 2003 2005 2007 2009 $Billion Source: Towers Watson Per-Case Cost of Defending Litigation 0 115 230 345 460 575 2001 2003 2005 3007 2009 $Thousand Source: Brestoff & Inmon
not to play! Or, Less Litigation.
Answer: Deep Learning All 47 Atari games The Asian game of Go (2016, 2017) Heads Up, No Limit Texas Hold em Poker (2017, 2x) 1. TRAINING DATA-Leverage labeled content (PACER!) 2. TEXT CLASSIFIER-Train a Deep Learning Model 3. SHIFT TO ENTERPRISE-Run on COMPANY content 4. EXAMINE RESULTS-Identify Risks for Early Warnings Why Deep Learning Is Suddenly Changing Your Life By Roger Parloff Fortune Magazine September 28, 2016
Sliding windows create number strings for words and keep the words in context The input word is highlighted in blue
WITH EXAMPLES FROM DISCRIMINATION LAWSUITS, A DEEP LEARNING SYSTEM CAN INFER DISCRIMINATION IN COMPANY DATA
T-SNE VISUALIZATION
Case-Types in PACER Nature of Suit (NOS) 160 Stockholders Suits NOS 190 Contracts NOS 370 Fraud NOS 410 Antitrust NOS 442 Civil Rights-Employment NOS 710 Fair Labor Standards Act
Using PACER, a Deep Learning system can be trained for a multiplicity of filters
Then you monitor your internal data -- to surface those risky needles in the haystack. You examine what the system finds as early warnings AI means Augmented Intelligence
When youre on the alert, you decide the scope of your investigation You decide which internal databases to access
HEAD LITIGATION OFF AT THE PASS. NIP IT IN THE BUD. DONT MANAGE LITIGATION AVOID IT THATS HOW DEEP LEARNING WILL CHANGE OUR PROFESSION. NOT PATENT PENDING. PATENTED.
Don't Manage Litigation,Prevent it! Use Deep Learning
Nick Brestoff, M.S., J.D. Inventor, Founder & CEO Nick.Brestoff@Intraspexion.co m 661-312-8518 (cell) 息 2017 Intraspexion Inc.

More Related Content

Don't Manage Litigation,Prevent it! Use Deep Learning

  • 4. LITIGATION-$160 BILLION PER YEAR (2001- 2010) EVEN WHEN A BUSINESS WINS, IT LOSES ABOUT $350,000 AND $400,000 PER CASE - SO THE ONLY WAY TO WIN IS U.S. Commercial Tort Costs 0. 45. 90. 135. 180. 225. 2001 2003 2005 2007 2009 $Billion Source: Towers Watson Per-Case Cost of Defending Litigation 0 115 230 345 460 575 2001 2003 2005 3007 2009 $Thousand Source: Brestoff & Inmon
  • 5. not to play! Or, Less Litigation.
  • 6. Answer: Deep Learning All 47 Atari games The Asian game of Go (2016, 2017) Heads Up, No Limit Texas Hold em Poker (2017, 2x) 1. TRAINING DATA-Leverage labeled content (PACER!) 2. TEXT CLASSIFIER-Train a Deep Learning Model 3. SHIFT TO ENTERPRISE-Run on COMPANY content 4. EXAMINE RESULTS-Identify Risks for Early Warnings Why Deep Learning Is Suddenly Changing Your Life By Roger Parloff Fortune Magazine September 28, 2016
  • 7. Sliding windows create number strings for words and keep the words in context The input word is highlighted in blue
  • 8. WITH EXAMPLES FROM DISCRIMINATION LAWSUITS, A DEEP LEARNING SYSTEM CAN INFER DISCRIMINATION IN COMPANY DATA
  • 10. Case-Types in PACER Nature of Suit (NOS) 160 Stockholders Suits NOS 190 Contracts NOS 370 Fraud NOS 410 Antitrust NOS 442 Civil Rights-Employment NOS 710 Fair Labor Standards Act
  • 11. Using PACER, a Deep Learning system can be trained for a multiplicity of filters
  • 12. Then you monitor your internal data -- to surface those risky needles in the haystack. You examine what the system finds as early warnings AI means Augmented Intelligence
  • 13. When youre on the alert, you decide the scope of your investigation You decide which internal databases to access
  • 14. HEAD LITIGATION OFF AT THE PASS. NIP IT IN THE BUD. DONT MANAGE LITIGATION AVOID IT THATS HOW DEEP LEARNING WILL CHANGE OUR PROFESSION. NOT PATENT PENDING. PATENTED.
  • 16. Nick Brestoff, M.S., J.D. Inventor, Founder & CEO Nick.Brestoff@Intraspexion.co m 661-312-8518 (cell) 息 2017 Intraspexion Inc.