Geometric Neural Phrase Pooling: Modeling the Spatial Co-occurrence of Neurons
This addresses a less-studied aspect of CNN performance for visual recognition, offering an incremental improvement with low computational overhead.
The paper tackled the problem of modeling spatial co-occurrence of neuron responses in CNNs by proposing Geometric Neural Phrase Pooling (GNPP), which groups neurons into phrases and punishes isolated responses, resulting in significant and consistent accuracy gains in image classification.
Deep Convolutional Neural Networks (CNNs) are playing important roles in state-of-the-art visual recognition. This paper focuses on modeling the spatial co-occurrence of neuron responses, which is less studied in the previous work. For this, we consider the neurons in the hidden layer as neural words, and construct a set of geometric neural phrases on top of them. The idea that grouping neural words into neural phrases is borrowed from the Bag-of-Visual-Words (BoVW) model. Next, the Geometric Neural Phrase Pooling (GNPP) algorithm is proposed to efficiently encode these neural phrases. GNPP acts as a new type of hidden layer, which punishes the isolated neuron responses after convolution, and can be inserted into a CNN model with little extra computational overhead. Experimental results show that GNPP produces significant and consistent accuracy gain in image classification.