Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationPariscript signature verification software
The software uses machine learning and neural networks to analyze signature features and compare them against reference signatures, enabling automated first-pass verification of ballots.
“The company is branched out into document processing systems for clients in a variety of industries, including banking and insurance. Voter ballots are just another kind of document, and the company's long history with handwriting analysis, came in handy for performing signature verification for election boards.”
Engineering & simulationPariscript software (new release)
A software update was released to enhance low-resolution images and adjust DPI differences, allowing the system to compare signatures of different qualities by correcting for resolution loss.
“Now, you can use machine learning to correct for that as well. So what we did earlier this year is have a new release of our software, which will actually enhance the image, but also adjust down and basically make a take an apples to orange issue where we've got 2 images at different DPIs and correct for that so that we can do a like for like.”
Engineering & simulationPariscript proprietary algorithms
The engineering team blends multiple orthogonal machine learning algorithms, including deep learning neural networks and computer vision techniques, to identify precise signature features and determine authenticity.
“So what they did is they blended a number of them, they like to use the word orthogonal, or, you know, it's as if you're using a number of different experts, each with their own opinions, and then on top of that, we have some voting algorithms that will then take all of these independent opinions and identify the most precise one or the most precise answer.”
Engineering & simulationPariscript software70-75% automation rate, 98% accuracy
The software is trained on specific DPI ranges (200-300) and uses a reject curve to balance automation rates against error rates, achieving 70-75% automation with 98% accuracy.
“So for instance, within our own data sets, we have, we can achieve about a 70 to 75% automation rate holding at about a 98% accuracy rate or 2% error. And, you know, that error rate can be, you know, halved. Of course, there's a corresponding reduction in the amount of automation…”
Running the companyPariscript software
The system processes ballots as a first-pass automation tool, automatically processing easy cases while flagging complex ones for human review, reducing the manual workload for election officials.
“So what it does ultimately is it allows the easy stuff, so to speak, to be able to be processed automatically, which reduces the amount of work that they have to do. And as we've seen with this recent election, They were still able to get through quite a few, you know, a large amount of ballots in a relatively short amount of time.”