Multiactivation Pooling Method in Convolutional Neural Networks for Image Recognition

Qi Zhao, Shuchang Lyu, Boxue Zhang, Wenquan Feng

Abstract: Convolutional neural networks (CNNs) are becoming more and more popular today. CNNs now have become a popular feature extractor applying to image processing, big data processing, fog computing, etc. CNNs usually consist of several basic units like convolutional unit, pooling unit, activation unit, and so on. In CNNs, conventional pooling methods refer to 2×2 max-pooling and average-pooling, which are applied after the convolutional or ReLU layers. In this paper, we propose a Multiactivation Pooling (MAP) Method to make the CNNs more accurate on classification tasks without increasing depth and trainable parameters. We add more convolutional layers before one pooling layer and expand the pooling region to 4×4, 8×8, 16×16, and even larger. When doing large-scale subsampling, we pick top-k activation, sum up them, and constrain them by a hyperparameter <span class="nowrap"><svg xmlns:xlink="http://www.w3.org/1999/xlink" xmlns="http://www.w3.org/2000/svg" style="vertical-align:-0.2063899pt" id="M1" height="6.34998pt" version="1.1" viewBox="-0.0498162 -6.14359 7.47218 6.34998" width="7.47218pt"><g transform="matrix(.013,0,0,-0.013,0,0)"><path id="g113-240" d="M548 455L523 469C505 448 490 439 461 439C416 439 374 447 316 447C144 447 23 310 23 161C23 47 89 -12 179 -12C312 -12 429 106 429 244C429 303 419 344 368 386L371 388C404 383 440 378 474 378C507 378 530 411 548 455ZM350 274C350 191 308 25 198 25C144 25 108 75 108 157C108 264 170 395 276 395C296 395 320 384 333 360C347 334 350 316 350 274Z"/></g></svg>.</span> We pick VGG, ALL-CNN, and DenseNets as our baseline models and evaluate our proposed MAP method on benchmark datasets: CIFAR-10, CIFAR-100, SVHN, and ImageNet. The classification results are competitive.