Chao Feng

Chao Feng, Jui Po Hung, Aishan Li et al.

Diabetic Retinopathy (DR) has become one of the leading causes of vision impairment in working-aged people and is a severe problem worldwide. However, most of the works ignored the ordinal information of labels. In this project, we propose a novel design MTCSNN, a Multi-task Clinical Siamese Neural Network for Diabetic Retinopathy severity prediction task. The novelty of this project is to utilize the ordinal information among labels and add a new regression task, which can help the model learn more discriminative feature embedding for fine-grained classification tasks. We perform comprehensive experiments over the RetinaMNIST, comparing MTCSNN with other models like ResNet-18, 34, 50. Our results indicate that MTCSNN outperforms the benchmark models in terms of AUC and accuracy on the test dataset.

Chao Feng, Ziyang Chen, Aleksander Holynski et al. · berkeley

We show that the GPS tags contained in photo metadata provide a useful control signal for image generation. We train GPS-to-image models and use them for tasks that require a fine-grained understanding of how images vary within a city. In particular, we train a diffusion model to generate images conditioned on both GPS and text. The learned model generates images that capture the distinctive appearance of different neighborhoods, parks, and landmarks. We also extract 3D models from 2D GPS-to-image models through score distillation sampling, using GPS conditioning to constrain the appearance of the reconstruction from each viewpoint. Our evaluations suggest that our GPS-conditioned models successfully learn to generate images that vary based on location, and that GPS conditioning improves estimated 3D structure.

Chao Feng, Yanze Zhang, Chenghao Zhang

This paper proposes a generative method to dynamically simulate users' short video watching journey for watch time prediction in short video recommendation. Unlike existing methods that rely on multimodal features for video content understanding, our method simulates users' sustained interest in watching short videos by learning collaborative information, using interest changes from existing positive and negative feedback videos and user interaction behaviors to implicitly model users' video watching journey. By segmenting videos based on duration and adopting a Transformer-like architecture, our method can capture sequential dependencies between segments while mitigating duration bias. Extensive experiments on industrial-scale and public datasets demonstrate that our method achieves state-of-the-art performance on watch time prediction tasks. The method has been deployed on Kuaishou Lite, achieving a significant improvement of +0.13\% in APP duration, and reaching an XAUC of 83\% for single video watch time prediction on industrial-scale streaming training sets, far exceeding other methods. The proposed method provides a scalable and effective solution for video recommendation through segment-level modeling and user engagement feedback.