Robust and Explainable Fine-Grained Visual Classification with Transfer Learning: A Dual-Carriageway Framework
This addresses a practical gap for developers in fine-grained visual classification by offering an automated, explainable method to decide training strategies, though it is incremental as it builds on existing transfer learning techniques.
The paper tackles the problem of choosing the optimal training strategy (training from scratch vs. fine-tuning) when a new dataset becomes available in fine-grained visual classification, presenting the Dual-Carriageway Framework (DCF) that automatically selects the best approach and provides explanations, with results showing fine-tuning outperforms training from scratch by up to 2.13% and 1.23% in mean accuracy on different datasets.
In the realm of practical fine-grained visual classification applications rooted in deep learning, a common scenario involves training a model using a pre-existing dataset. Subsequently, a new dataset becomes available, prompting the desire to make a pivotal decision for achieving enhanced and leveraged inference performance on both sides: Should one opt to train datasets from scratch or fine-tune the model trained on the initial dataset using the newly released dataset? The existing literature reveals a lack of methods to systematically determine the optimal training strategy, necessitating explainability. To this end, we present an automatic best-suit training solution searching framework, the Dual-Carriageway Framework (DCF), to fill this gap. DCF benefits from the design of a dual-direction search (starting from the pre-existing or the newly released dataset) where five different training settings are enforced. In addition, DCF is not only capable of figuring out the optimal training strategy with the capability of avoiding overfitting but also yields built-in quantitative and visual explanations derived from the actual input and weights of the trained model. We validated DCF's effectiveness through experiments with three convolutional neural networks (ResNet18, ResNet34 and Inception-v3) on two temporally continued commercial product datasets. Results showed fine-tuning pathways outperformed training-from-scratch ones by up to 2.13% and 1.23% on the pre-existing and new datasets, respectively, in terms of mean accuracy. Furthermore, DCF identified reflection padding as the superior padding method, enhancing testing accuracy by 3.72% on average. This framework stands out for its potential to guide the development of robust and explainable AI solutions in fine-grained visual classification tasks.