TY - JOUR
T1 - Beyond Semantic Attributes: Discrete Latent Attributes Learning for Zero-Shot Recognition
AU - Qin, Jie
AU - Wang, Yunhong
AU - Liu, Li
AU - Chen, Jiaxin
AU - Shao, Ling
PY - 2016/9/21
Y1 - 2016/9/21
N2 - In this letter, we propose a novel approach for learning semantics-driven attributes, which are discriminative for zero-shot visual recognition. Latent attributes are derived in a principled manner, aiming at maintaining class-level semantic relatedness and attribute-wise balancedness. Unlike existing methods that binarize learned real-valued attributes via a quantization stage, we directly learn the optimal binary attributes by effectively addressing a discrete optimization problem. Particularly, we propose a class-wise discrete descent algorithm, based on which latent attributes of each class are learned iteratively. Moreover, we propose to simultaneously predict multiple attributes from low-level features via multioutput neural networks (MONN), which can model intrinsic correlation among attributes and make prediction more tractable. Extensive experiments on two standard datasets clearly demonstrate the superiority of our method over the state-of-the-arts.
AB - In this letter, we propose a novel approach for learning semantics-driven attributes, which are discriminative for zero-shot visual recognition. Latent attributes are derived in a principled manner, aiming at maintaining class-level semantic relatedness and attribute-wise balancedness. Unlike existing methods that binarize learned real-valued attributes via a quantization stage, we directly learn the optimal binary attributes by effectively addressing a discrete optimization problem. Particularly, we propose a class-wise discrete descent algorithm, based on which latent attributes of each class are learned iteratively. Moreover, we propose to simultaneously predict multiple attributes from low-level features via multioutput neural networks (MONN), which can model intrinsic correlation among attributes and make prediction more tractable. Extensive experiments on two standard datasets clearly demonstrate the superiority of our method over the state-of-the-arts.
KW - Class-level attribute learning
KW - discrete optimization
KW - multioutput neural networks
KW - zero-shot learning
U2 - 10.1109/LSP.2016.2612247
DO - 10.1109/LSP.2016.2612247
M3 - Article
VL - 23
SP - 1667
EP - 1671
JO - IEEE Signal Processing Letters
JF - IEEE Signal Processing Letters
SN - 1070-9908
IS - 11
ER -