Finalizing Results
Rohan K -
Welcome back! Last week I discussed my goal for the end result. I mentioned I wanted 90-95% accuracy for sentences found, ideally greater than 95% so that most of the time it would find all 20 sentences. After running my most recent model, I achieved 93.98% accuracy! We are very close to the end which is exciting. I am currently running an optimization to fine-tune the parameters a bit so we can hopefully get above 95%. This search will take about two days, so as it runs I am working on the presentation portion. A challenge will be explaining the project in just 12 minutes. I will have to be very concise and resourceful of images to do some of the explaining. I’m planning on including a ton of images of audio to demonstrate how the model works, so I need to confirm with my advisor if I can use anonymous patient data, or if I have to create my own example recordings. Patient confidentiality is important when handling clinical data.
As I finish building my model, I’m thinking about the finishing touches. I need to clean up the code so that it’s readable to anyone using it, and so that it’s customizable to any number of sentences. I need to test my model on healthy speech to see what the accuracy is. It’s expected that it won’t perform well because it’s tuned specific to LD/tremor. I also need to test it on another set of LD/tremor recordings that I haven’t looked at yet, to see if my model is generalizable. These final tasks are not necessary, but would be nice to have. Hopefully I can give you some of these numbers in the next two weeks.
Comments:
All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.