4.4. Discussion 4 51 Figure 4.8: Results of experiment 4. The model was trained on naturalistic stimuli with a customized phosphene mapping. a) Reconstruction performance for the different phosphene resolutions (AUC: area under the receiveroperator curve) b) Visualization of the phosphene coverage for each resolution (left: 650 phosphenes, middle: 488 phosphenes, right: 325 phosphenes). c) Validation examples for the training condition with 650 phosphenes. 4.4. Discussion In this paper we present and evaluate a novel DL approach for end-to-end optimization of prosthetic vision. Below we provide a general discussion of the proposed method and the results of our validation experiments, reflecting on the earlier hypothesized automated and tailored optimization abilities. Furthermore, we list some of the limitations of the current study and provide directions for future research. 4.4.1. Automated optimization Our end-to-end model is based on an autoencoder architecture, and aims to make use of their well-described ability to efficiently encode information into a low-dimensional latent representation (Bengio et al., 2013). Instead of optimizing image preprocessing as an isolated operation, our approach is designed to automatically optimize the entire process of phosphene generation for a given task. The results from experiment 1 demonstrate that the model successfully converges to an optimal encoding strategy for a latent representation that consisted of a 32 × 32 binary simulated phosphene pattern. The model achieved adequate reconstruction performance, as indicated by the low MSE of 0.018 on the validation dataset. Note, that in the unconstrained setting, the model merely
RkJQdWJsaXNoZXIy MTk4NDMw