Language Learning Using Speech To Image Retrieval

Abstract

Humans learn language by interaction with their environment and listening to other humans. It should also be possible for computational models to learn language directly from speech but so far most approaches require text. We improve on existing neural network approaches to create visually grounded embeddings for spoken utterances. Using a combination of a multi-layer GRU, importance sampling, cyclic learning rates, ensembling and vectorial self-attention our results show a remarkable increase in image-caption retrieval performance over previous work. Furthermore, we investigate which layers in the model learn to recognise words in the input. We find that deeper network layers are better at encoding word presence, although the final layer has slightly lower performance. This shows that our visually grounded sentence encoder learns to recognise words from the input even though it is not explicitly trained for word recognition.

Stats

• citations17
• S2 citations—
• github stars0
• HF likes0
• heat score9.41
• arxiv keymerkx2019language

Related papers

• Leveraging Pretrained Image-text Models For Improving Audio-visual Learning (2023)0.00
• Semantic Speech Retrieval With A Visually Grounded Model Of Untranscribed Speech (2017)10.61
• Learning Word-like Units From Joint Audio-visual Analysis (2017)12.33
• Large-scale Representation Learning From Visually Grounded Untranscribed Speech (2019)10.48
• Catplayinginthesnow: Impact Of Prior Segmentation On A Model Of Visually Grounded Speech (2020)4.52
• Vision As An Interlingua: Learning Multilingual Semantic Embeddings Of Untranscribed Speech (2018)10.74
• Jointly Discovering Visual Objects And Spoken Words From Raw Sensory Input (2018)14.27
• Symbolic Inductive Bias For Visually Grounded Learning Of Spoken Language (2018)5.24