Hyperbolic Audio-visual Zero-shot Learning

Audio-visual zero-shot learning aims to classify samples consisting of a pair of corresponding audio and video sequences from classes that are not present during training. An analysis of the audio-visual data reveals a large degree of hyperbolicity, indicating the potential benefit of using a hyperbolic transformation to achieve curvature-aware geometric learning, with the aim of exploring more complex hierarchical data structures for this task. The proposed approach employs a novel loss function that incorporates cross-modality alignment between video and audio features in the hyperbolic space. Additionally, we explore the use of multiple adaptive curvatures for hyperbolic projections. The experimental results on this very challenging task demonstrate that our proposed hyperbolic approach for zero-shot learning outperforms the SOTA method on three datasets: VGGSound-GZSL, UCF-GZSL, and ActivityNet-GZSL achieving a harmonic mean (HM) improvement of around 3.0%, 7.0%, and 5.3%, respectively.

PDF Abstract ICCV 2023 PDF ICCV 2023 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
GZSL Video Classification ActivityNet-GZSL (cls) Hyper-multiple HM 15.25 # 2
ZSL 10.39 # 2
GZSL Video Classification ActivityNet-GZSL(main) Hyper-multiple HM 12.65 # 2
ZSL 9.50 # 2
GZSL Video Classification UCF-GZSL (cls) Hyper-multiple HM 48.30 # 3
ZSL 52.11 # 2
GZSL Video Classification UCF-GZSL(main) Hyper-multiple HM 29.32 # 3
ZSL 22.24 # 3
GZSL Video Classification VGGSound-GZSL (cls) Hyper-multiple HM 8.67 # 3
ZSL 7.31 # 3
GZSL Video Classification VGGSound-GZSL(main) Hyper-multiple HM 9.32 # 2
ZSL 7.97 # 2


No methods listed for this paper. Add relevant methods here