 Hey everyone, I'm Sartha, a freshman at Purdue studying computer science. I'm an undergraduate researcher working with Professor Heinz at the Heinz Lab, and today I'm going to be presenting an individualized adaptive deep learning based hearing aid. First off, let's discuss the problem. To do this, I'm going to introduce two categories, one for people with mild hearing loss and the other for people with moderate to profound hearing loss. For category one, the crucial limitations lie in the limited benefits offered from using hearing aids and the low accessibility of professional help. Additionally, hearing aids are expensive and based on a one size fits all model. For category two, the key limitations are more technological, especially including the limited human machine interface and the individualized fitting and diagnostic procedures. Similar to category one, they also suffer from the cost and low availability of professional help. Okay, so now for the solution. What is Indivir and how does it work? So we utilized a convolutional autoencoder modified with the bi-LSCM layer to factor in the context of the audio sample. The network is trained separately for each patient and it can be referenced in real time on any smartphone device. To achieve this, we converted the original model by Facebook research from PyTorch to TensorFlow Lite. Now let's talk about how this model is trained for a given patient. Indivir follows a fully remote two-step process for the same. The model is initialized using weights from denoiser. Step one focuses on fine tuning the model for the patient. If patients with similar hearing profiles exist, then those models are utilized here. Otherwise, a model is trained from scratch using a speech-autometry-based data collection process. Step two focuses on using reinforcement learning to align the model with the user's individual preferences and needs. Feedback from the patient is taken regularly and the model is adjusted accordingly. To carry all of this out, we have developed both software and hardware prototypes. The approximate cost is around $70 for both ears and the software can be done on any smartphone device. Finally, thank you for your time and attention.