# AI proposal

跟signoff有關 需要100% accuracy的task，不能仰賴DL model

## Predictive Analysis for Early Detection:
目的不是要取代simulation，是方便我們可以在早期就可以得到預警當作參考

1. Critical Path Identification: Use AI to analyze historical data and identify critical paths in the design that are most likely to fail LVS, RC extraction, or electromigration checks. This allows designers to focus their efforts on the most problematic areas early in the process.

2. Anomaly Detection: Implement AI to detect unusual patterns or anomalies in layout and schematic data that could indicate potential issues. This preemptive step can catch errors that might otherwise go unnoticed until later stages.

## Intelligent Prioritization and Triage:

1. Issue Prioritization: Develop AI models that can prioritize detected issues based on their potential impact on the final design. This helps streamline the signoff process by focusing human efforts on the most critical problems first.

2. Automated Triage: Use AI to classify and triage issues, suggesting possible root causes and potential fixes. This reduces the time spent on diagnosing problems and accelerates the resolution process.

## Electromigration Risk Assessment:

1. Predictive Modeling: Utilize AI to create predictive models for electromigration based on historical failure data. These models can identify potential high-risk areas in the design, allowing for proactive adjustments.