Multimodal Fusion and Learning with Uncertain Features Applied to Audiovisual Speech Recognition | IEEE Conference Publication | IEEE Xplore