Hierarchical classification predicts labels across multiple levels of a taxonomy, e.g., from coarse-level \textit{Bird} to mid-level \textit{Hummingbird} to fine-level \textit{Green hermit}, allowing flexible recognition under varying visual conditions. It is commonly framed as multiple single-level tasks, but each level may rely on different visual cues. Distinguishing \textit{Bird} from \textit{Plant} relies on {\it global features} like {\it feathers} or {\it leaves}, while separating \textit{Anna's hummingbird} from \textit{Green hermit} requires {\it local details} such as {\it head coloration}. Prior methods improve accuracy using external semantic supervision, but such statistical learning criteria fail to ensure consistent visual grounding at test time, resulting in incorrect hierarchical classification. We propose, for the first time, to enforce \textit{internal visual consistency} by aligning fine-to-coarse predictions through intra-image segmentation. Our method outperforms zero-shot CLIP and state-of-the-art baselines on hierarchical classification benchmarks, achieving both higher accuracy and more consistent predictions. It also improves internal image segmentation without requiring pixel-level annotations.

Given a set of images, our goal is to map each image to a point in a feature space such that, not only point proximity indicates visual similarity, but where it is located directly encodes how prototypical the image is according to the dataset. Our key insight is to perform unsupervised feature learning in hyperbolic instead of Euclidean space, where the distance between points still reflects image similarity, yet we gain additional capacity for representing prototypicality with the location of the point: The closer it is to the origin, the more prototypical it is. The latter property is simply emergent from optimizing the metric learning objective: The image similar to many training instances is best placed at the center of corresponding points in Euclidean space, but closer to the origin in hyperbolic space. We propose an unsupervised feature learning algorithm in Hyperbolic space with sphere pACKing. HACK first generates uniformly packed particles in the Poincaré ball of hyperbolic space and then assigns each image uniquely to a particle. With our feature mapper simply trained to spread out training instances in hyperbolic space, we observe that images move closer to the origin with congealing - a warping process that aligns all the images and makes them appear more common and similar to each other, validating our idea of unsupervised prototypicality discovery. We demonstrate that our data-driven prototypicality provides an easy and superior unsupervised instance selection to reduce sample complexity, increase model generalization with atypical instances and robustness with typical ones.

Achieving the pathological type diagnosis of pulmonary nodules on chest CT is a critical step in the early detection of lung cancer and treatment of patients. Based on a small and unbalanced self-constructed dataset, we achieved intelligent diagnosis of five pathological types including adenocarcinoma, squamous cell carcinoma, small cell carcinoma, inflammatory and other benign diseases for the first time. In order to reduce the dependence of deep convolutional neural network (DCNN) on a large amount of training data, a reverse adversarial classification network (RACN) was proposed based on semi-supervised learning, which consists of a reverse generative adversarial network (RGAN) for unsupervised regression and a supervised classification network (CN). In RGAN, five specific normal distributions P with different means and variances were assigned to represent the five pathological types, and then a special regression task was designed by mapping pulmonary nodules to the random sampling Z of P. The input of generator in RGAN is set to 3D nodule volume data, the inputs of discriminator are set to Z and the output of generator. The regression task enables RGAN to extract specific features, which will be deeply integrate into CN to improve the classification performance. Experiments showed that the average sensitivity of RACN in detecting malignant nodules was 0.6525, where the sensitivity of adenocarcinoma, small cell carcinoma and squamous cell carcinoma was 0.8426, 0.5604 and 0.5543. Besides, the RACN can achieve 93.21% accuracy for diagnosing malignant nodules on the public LIDC-IDRI dataset, obtaining the state-of-the-art results.