AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Unsupervised Learning of Spoken Language with Visual Context

David Harwath, Antonio Torralba, James Glass

NeurIPS 2016 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
truncate audio captions to 20 s instead of 10 s (ablation)audio-image model · Places audio captionsrecall: .056 -> .066Improved
max instead of mean pooling over audio caption (ablation)audio-image model · Places audio captionsrecallNo clear effect
L2-normalize only the acoustic embeddings (ablation)audio-image model · Places audio captionsrecallImproved
also L2-normalize image embeddings (cosine similarity instead of dot product) (ablation)audio-image model · Places audio captionsrecallHurt
embedding dimension d = 1024 (ablation)audio-image model · Places audio captionsrecallImproved
use first conv layer features versus MFCC39 baselineaudio-image model · TIMIT keyword spottingEER: 0.127 -> 0.049Improved
compare to best prior unsupervised [12]audio-image model · TIMIT keyword spottingP@N: 0.63 -> 0.62No clear effect
compare to best prior unsupervised [12]audio-image model · TIMIT keyword spottingEER: 0.169 -> 0.049Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy