Binary Stochastic Representations for Large Multi-class Classification
This addresses the inference efficiency issue for applications like image or document tagging in social networks, offering an incremental improvement over existing binary code methods by eliminating the need for heuristic-based code assignment.
The paper tackles the problem of high inference complexity in large multi-class classification by proposing Deep Stochastic Neural Codes (DSNC), an end-to-end model that simultaneously learns binary code assignments and input mappings, achieving sublinear inference complexity without prior tuning and showing effectiveness on various datasets.
Classification with a large number of classes is a key problem in machine learning and corresponds to many real-world applications like tagging of images or textual documents in social networks. If one-vs-all methods usually reach top performance in this context, these approaches suffer from a high inference complexity, linear w.r.t the number of categories. Different models based on the notion of binary codes have been proposed to overcome this limitation, achieving in a sublinear inference complexity. But they a priori need to decide which binary code to associate to which category before learning using more or less complex heuristics. We propose a new end-to-end model which aims at simultaneously learning to associate binary codes with categories, but also learning to map inputs to binary codes. This approach called Deep Stochastic Neural Codes (DSNC) keeps the sublinear inference complexity but do not need any a priori tuning. Experimental results on different datasets show the effectiveness of the approach w.r.t baseline methods.