Deep Learning to Ternary Hash Codes by Continuation
This work addresses the need for efficient image retrieval systems by improving ternary code generation, though it appears incremental as it builds on existing deep learning and hashing techniques.
The paper tackles the problem of generating high-quality ternary hash codes for image retrieval by proposing a method to jointly learn features and codes using a smoothed function that evolves via continuation, resulting in higher retrieval accuracy.
Recently, it has been observed that {0,1,-1}-ternary codes which are simply generated from deep features by hard thresholding, tend to outperform {-1,1}-binary codes in image retrieval. To obtain better ternary codes, we for the first time propose to jointly learn the features with the codes by appending a smoothed function to the networks. During training, the function could evolve into a non-smoothed ternary function by a continuation method. The method circumvents the difficulty of directly training discrete functions and reduces the quantization errors of ternary codes. Experiments show that the generated codes indeed could achieve higher retrieval accuracy.