Summary of Automated Neural Patent Landscaping in the Small Data Regime, by Tisa Islam Erana and Mark A. Finlayson
Automated Neural Patent Landscaping in the Small Data Regime
by Tisa Islam Erana, Mark A. Finlayson
First submitted to arxiv on: 10 Jul 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents an automated neural patent landscaping system that improves performance on difficult examples while requiring less training data. The system demonstrates significant advancements over previous methods, achieving 0.69 F1 score on ‘hard’ examples and 0.75 F1 score with as few as 24 labeled examples. The authors also introduce a higher-quality training data generation procedure by merging the “seed/anti-seed” approach with active learning to collect difficult labeled examples near the decision boundary. This new dataset of labeled AI patents is released for others to build upon, along with baseline comparisons and code. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper describes an automated way to create a map of all patents related to a specific technology area. This is important because it helps us understand intellectual property better. The system is faster and more accurate than previous methods and only needs a small amount of labeled examples to train. It also generates better training data by combining two existing approaches with active learning. This new approach is released for others to use. |
Keywords
» Artificial intelligence » Active learning » F1 score
