More Like this

1. An Empirical Analysis of Diffusion, Autoencoders, and Adversarial Deep Learning Models for Predicting Dementia Using High-Fidelity MRI

   https://doi.org/10.1109/ACCESS.2024.3354724  

   Gajjar, Pranshav; Garg, Manav; Desai, Shivani; Chhinkaniwala, Hitesh; Sanghvi, Harshal A; Patel, Riki H; Gupta, Shailesh; Pandya, Abhijit S (January 2024, IEEE Access)

   Saif, Mehrdad (Ed.)

   This study explores cutting-edge computational technologies and intelligent methods to create realistic synthetic data, focusing on dementia-centric Magnetic Resonance Imaging (MRI) scans related to Alzheimer’s and Parkinson’s diseases. The research delves into Generative Adversarial Networks (GANs), Variational Autoencoders, and Diffusion Models, comparing their efficacy in generating synthetic MRI scans. Using datasets from Alzheimer’s and Parkinson’s patients, the study reveals intriguing findings. In the Alzheimer dataset, diffusion models produced non-dementia images with the lowest Frechet Inception Distance (FID) score at 92.46, while data-efficient GANs excelled in generating dementia images with an FID score of 178.53. In the Parkinson dataset, data-efficient GANs achieved remarkable FID scores of 102.71 for dementia images and 129.77 for non-dementia images. The study also introduces a novel aspect by incorporating a classification study, validating the generative metrics. DenseNets, a deep learning architecture, exhibited superior performance in disease detection compared to ResNets. Training both models on images generated by diffusion models further improved results, with DenseNet achieving accuracies of 80.84% and 92.42% in Alzheimer’s and Parkinson’s disease detection, respectively. The research not only presents innovative generative architectures but also emphasizes the importance of classification metrics, providing valuable insights into the synthesis and detection of neurodegenerative diseases through advanced computational techniques. 

   more » « less
2. Patch Diffusion: Faster and More Data-Efficient Training of Diffusion Models

   Wang, Zhendong; Jiang, Yifan; Zheng, Huangjie; Wang, Peihao; He, Pengcheng; Wang, Zhangyang; Chen, Weizhu; Zhou, Mingyuan (December 2023, Neural Information Processing Systems)

   Diffusion models are powerful, but they require a lot of time and data to train. We propose Patch Diffusion, a generic patch-wise training framework, to significantly reduce the training time costs while improving data efficiency, which thus helps democratize diffusion model training to broader users. At the core of our innovations is a new conditional score function at the patch level, where the patch location in the original image is included as additional coordinate channels, while the patch size is randomized and diversified throughout training to encode the cross-region dependency at multiple scales. Sampling with our method is as easy as in the original diffusion model. Through Patch Diffusion, we could achieve ≥2× faster training, while maintaining comparable or better generation quality. Patch Diffusion meanwhile improves the performance of diffusion models trained on relatively small datasets, e.g., as few as 5,000 images to train from scratch. We achieve outstanding FID scores in line with state-of-the-art benchmarks: 1.77 on CelebA-64×64, 1.93 on AFHQv2-Wild-64×64, and 2.72 on ImageNet-256×256. We share our code and pre-trained models in GitHub. 

   more » « less
3. Progressive distillation induces an implicit curriculum

   Panigrahi, A; Liu, B; Malladi, S; Risteski, A; Goel, S (April 2025, International Conference on Learning Representations (ICLR), 2025)

   Knowledge distillation leverages a teacher model to improve the training of a student model. A persistent challenge is that a better teacher does not always yield a better student, to which a common mitigation is to use additional supervision from several “intermediate” teachers. One empirically validated variant of this principle is progressive distillation, where the student learns from successive intermediate checkpoints of the teacher. Using sparse parity as a sandbox, we identify an implicit curriculum as one mechanism through which progressive distillation accelerates the student’s learning. This curriculum is available only through the intermediate checkpoints but not the final converged one, and imparts both empirical acceleration and a provable sample complexity benefit to the student. We then extend our investigation to Transformers trained on probabilistic context-free grammars (PCFGs) and real-world pre-training datasets (Wikipedia and Books). Through probing the teacher model, we identify an analogous implicit curriculum where the model progressively learns features that capture longer context. Our theoretical and empirical findings on sparse parity, complemented by empirical observations on more complex tasks, highlight the benefit of progressive distillation via implicit curriculum across setups. 

   more » « less
4. SeqPATE: Differentially Private Text Generation via Knowledge Distillation

   Tian, Zhiliang; Zhao, Yingxiu; Huang, Ziyue; Wang, Yu-Xiang; Zhang, Nevin; He, He (December 2022, Advances in neural information processing systems)

   Protecting the privacy of user data is crucial for text generation models, which can leak sensitive information during generation. Differentially private (DP) learning methods provide guarantees against identifying the existence of a training sample from model outputs. PATE is a recent DP learning algorithm that achieves high utility with strong privacy protection on training samples. However, text generation models output tokens sequentially in a large output space; the classic PATE algorithm is not customized for this setting. Furthermore, PATE works well to protect sample-level privacy, but is not designed to protect phrases in samples. In this paper, we propose SeqPATE, an extension of PATE to text generation that protects the privacy of individual training samples and sensitive phrases in training data. To adapt PATE to text generation, we generate pseudo-contexts and reduce the sequence generation problem to a next-word prediction problem. To handle the large output space, we propose a candidate filtering strategy to dynamically reduce the output space, and refine the teacher aggregation of PATE to avoid low agreement due to voting for a large number of candidates. To further reduce privacy losses, we use knowledge distillation to reduce the number of teacher queries. The experiments verify the effectiveness of SeqPATE in protecting both training samples and sensitive phrases. 

   more » « less
5. Cooperative Knowledge Distillation: A Learner Agnostic Approach

   https://doi.org/10.1609/aaai.v38i13.29322  

   Livanos, Michael; Davidson, Ian; Wong, Stephen (March 2024, Proceedings of the AAAI Conference on Artificial Intelligence)

   Knowledge distillation is a simple but powerful way to transfer knowledge between a teacher model to a student model. Existing work suffers from at least one of the following key limitations in terms of direction and scope of transfer which restrict its use: all knowledge is transferred from teacher to student regardless of whether or not that knowledge is useful, the student is the only one learning in this exchange, and typically distillation transfers knowledge only from a single teacher to a single student. We formulate a novel form of knowledge distillation in which many models can act as both students and teachers which we call cooperative distillation. The models cooperate as follows: a model (the student) identifies specific deficiencies in it's performance and searches for another model (the teacher) who encodes learned knowledge into instructional virtual instances via counterfactual instance generation. Because different models may have different strengths and weaknesses, all models can act as either students or teachers (cooperation) when appropriate and only distill knowledge in areas specific to their strengths (focus). Since counterfactuals as a paradigm are not tied to any specific algorithm, we can use this method to distill knowledge between learners of different architectures, algorithms, and even feature spaces. We demonstrate our approach not only outperforms baselines such as transfer learning, self-supervised learning, and multiple knowledge distillation algorithms on several datasets, but it can also be used in settings where the aforementioned techniques cannot. 

   more » « less