Efficient Adaptive Ensembling for Image Classification

In recent times, with the exception of sporadic cases, the trend in Computer Vision is to achieve minor improvements compared to considerable increases in complexity. To reverse this trend, we propose a novel method to boost image classification performances without increasing complexity. To this end, we revisited ensembling, a powerful approach, often not used properly due to its more complex nature and the training time, so as to make it feasible through a specific design choice. First, we trained two EfficientNet-b0 end-to-end models (known to be the architecture with the best overall accuracy/complexity trade-off for image classification) on disjoint subsets of data (i.e. bagging). Then, we made an efficient adaptive ensemble by performing fine-tuning of a trainable combination layer. In this way, we were able to outperform the state-of-the-art by an average of 0.5$\%$ on the accuracy, with restrained complexity both in terms of the number of parameters (by 5-60 times), and the FLoating point Operations Per Second (FLOPS) by 10-100 times on several major benchmark datasets.

PDF Abstract Expert Systems, 2023 PDF Expert Systems, 2023 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Image Classification CIFAR-10 efficient adaptive ensembling Accuracy 99.612 # 1
Image Classification CIFAR-100 efficient adaptive ensembling Accuracy 96.808 # 1
Image Classification CINIC-10 efficient adaptive ensembling Accuracy 95.064 # 2
Image Classification Food-101 efficient adaptive ensembling Accuracy 96.879 # 1
Image Classification Oxford-IIIT Pets efficient adaptive ensembling Accuracy 98.22 # 1
Image Classification Stanford Cars efficient adaptive ensembling Accuracy 99.847 # 1

Methods