Wei Huang

Yinda Chen, Che Liu, Wei Huang et al.

Vision-Language Pretraining (VLP) has demonstrated remarkable capabilities in learning visual representations from textual descriptions of images without annotations. Yet, effective VLP demands large-scale image-text pairs, a resource that suffers scarcity in the medical domain. Moreover, conventional VLP is limited to 2D images while medical images encompass diverse modalities, often in 3D, making the learning process more challenging. To address these challenges, we present Generative Text-Guided 3D Vision-Language Pretraining for Unified Medical Image Segmentation (GTGM), a framework that extends of VLP to 3D medical images without relying on paired textual descriptions. Specifically, GTGM utilizes large language models (LLM) to generate medical-style text from 3D medical images. This synthetic text is then used to supervise 3D visual representation learning. Furthermore, a negative-free contrastive learning objective strategy is introduced to cultivate consistent visual representations between augmented 3D medical image patches, which effectively mitigates the biases associated with strict positive-negative sample pairings. We evaluate GTGM on three imaging modalities - Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and electron microscopy (EM) over 13 datasets. GTGM's superior performance across various medical image segmentation tasks underscores its effectiveness and versatility, by enabling VLP extension into 3D medical imagery while bypassing the need for paired text.

Chen Zhu, Renkun Ni, Zheng Xu et al.

Innovations in neural architectures have fostered significant breakthroughs in language modeling and computer vision. Unfortunately, novel architectures often result in challenging hyper-parameter choices and training instability if the network parameters are not properly initialized. A number of architecture-specific initialization schemes have been proposed, but these schemes are not always portable to new architectures. This paper presents GradInit, an automated and architecture agnostic method for initializing neural networks. GradInit is based on a simple heuristic; the norm of each network layer is adjusted so that a single step of SGD or Adam with prescribed hyperparameters results in the smallest possible loss value. This adjustment is done by introducing a scalar multiplier variable in front of each parameter block, and then optimizing these variables using a simple numerical scheme. GradInit accelerates the convergence and test performance of many convolutional architectures, both with or without skip connections, and even without normalization layers. It also improves the stability of the original Transformer architecture for machine translation, enabling training it without learning rate warmup using either Adam or SGD under a wide range of learning rates and momentum coefficients. Code is available at https://github.com/zhuchen03/gradinit.

Wei Huang, Qinying Gu, Nanyang Ye

Offline reinforcement learning (RL) enables policy training solely on pre-collected data, avoiding direct environment interaction - a crucial benefit for energy-constrained embodied AI applications. Although Artificial Neural Networks (ANN)-based methods perform well in offline RL, their high computational and energy demands motivate exploration of more efficient alternatives. Spiking Neural Networks (SNNs) show promise for such tasks, given their low power consumption. In this work, we introduce DSFormer, the first spike-driven transformer model designed to tackle offline RL via sequence modeling. Unlike existing SNN transformers focused on spatial dimensions for vision tasks, we develop Temporal Spiking Self-Attention (TSSA) and Positional Spiking Self-Attention (PSSA) in DSFormer to capture the temporal and positional dependencies essential for sequence modeling in RL. Additionally, we propose Progressive Threshold-dependent Batch Normalization (PTBN), which combines the benefits of LayerNorm and BatchNorm to preserve temporal dependencies while maintaining the spiking nature of SNNs. Comprehensive results in the D4RL benchmark show DSFormer's superiority over both SNN and ANN counterparts, achieving 78.4% energy savings, highlighting DSFormer's advantages not only in energy efficiency but also in competitive performance. Code and models are public at https://wei-nijuan.github.io/DecisionSpikeFormer.

Haonan Wang, Wei Huang, Ziwei Wu et al.

Active learning theories and methods have been extensively studied in classical statistical learning settings. However, deep active learning, i.e., active learning with deep learning models, is usually based on empirical criteria without solid theoretical justification, thus suffering from heavy doubts when some of those fail to provide benefits in real applications. In this paper, by exploring the connection between the generalization performance and the training dynamics, we propose a theory-driven deep active learning method (dynamicAL) which selects samples to maximize training dynamics. In particular, we prove that the convergence speed of training and the generalization performance are positively correlated under the ultra-wide condition and show that maximizing the training dynamics leads to better generalization performance. Furthermore, to scale up to large deep neural networks and data sets, we introduce two relaxations for the subset selection problem and reduce the time complexity from polynomial to constant. Empirical results show that dynamicAL not only outperforms the other baselines consistently but also scales well on large deep learning models. We hope our work would inspire more attempts on bridging the theoretical findings of deep networks and practical impacts of deep active learning in real applications.

Hongyu Liu, Ziyu Wan, Wei Huang et al.

We propose PD-GAN, a probabilistic diverse GAN for image inpainting. Given an input image with arbitrary hole regions, PD-GAN produces multiple inpainting results with diverse and visually realistic content. Our PD-GAN is built upon a vanilla GAN which generates images based on random noise. During image generation, we modulate deep features of input random noise from coarse-to-fine by injecting an initially restored image and the hole regions in multiple scales. We argue that during hole filling, the pixels near the hole boundary should be more deterministic (i.e., with higher probability trusting the context and initially restored image to create natural inpainting boundary), while those pixels lie in the center of the hole should enjoy more degrees of freedom (i.e., more likely to depend on the random noise for enhancing diversity). To this end, we propose spatially probabilistic diversity normalization (SPDNorm) inside the modulation to model the probability of generating a pixel conditioned on the context information. SPDNorm dynamically balances the realism and diversity inside the hole region, making the generated content more diverse towards the hole center and resemble neighboring image content more towards the hole boundary. Meanwhile, we propose a perceptual diversity loss to further empower PD-GAN for diverse content generation. Experiments on benchmark datasets including CelebA-HQ, Places2 and Paris Street View indicate that PD-GAN is effective for diverse and visually realistic image restoration.

Hongyu Liu, Ziyu Wan, Wei Huang et al.

User-intended visual content fills the hole regions of an input image in the image editing scenario. The coarse low-level inputs, which typically consist of sparse sketch lines and color dots, convey user intentions for content creation (\ie, free-form editing). While existing methods combine an input image and these low-level controls for CNN inputs, the corresponding feature representations are not sufficient to convey user intentions, leading to unfaithfully generated content. In this paper, we propose DeFLOCNet which relies on a deep encoder-decoder CNN to retain the guidance of these controls in the deep feature representations. In each skip-connection layer, we design a structure generation block. Instead of attaching low-level controls to an input image, we inject these controls directly into each structure generation block for sketch line refinement and color propagation in the CNN feature space. We then concatenate the modulated features with the original decoder features for structure generation. Meanwhile, DeFLOCNet involves another decoder branch for texture generation and detail enhancement. Both structures and textures are rendered in the decoder, leading to user-intended editing results. Experiments on benchmarks demonstrate that DeFLOCNet effectively transforms different user intentions to create visually pleasing content.

Chaorong Li, Malu Zhang, Wei Huang et al.

Deep Convolutional Neural Network (DCNN) and Transformer have achieved remarkable successes in image recognition. However, their performance in fine-grained image recognition is still difficult to meet the requirements of actual needs. This paper proposes a Sequence Random Network (SRN) to enhance the performance of DCNN. The output of DCNN is one-dimensional features. This one-dimensional feature abstractly represents image information, but it does not express well the detailed information of image. To address this issue, we use the proposed SRN which composed of BiLSTM and several Tanh-Dropout blocks (called BiLSTM-TDN), to further process DCNN one-dimensional features for highlighting the detail information of image. After the feature transform by BiLSTM-TDN, the recognition performance has been greatly improved. We conducted the experiments on six fine-grained image datasets. Except for FGVC-Aircraft, the accuracies of the proposed methods on the other datasets exceeded 99%. Experimental results show that BiLSTM-TDN is far superior to the existing state-of-the-art methods. In addition to DCNN, BiLSTM-TDN can also be extended to other models, such as Transformer.

Hongyu Liu, Bin Jiang, Yibing Song et al.

Deep encoder-decoder based CNNs have advanced image inpainting methods for hole filling. While existing methods recover structures and textures step-by-step in the hole regions, they typically use two encoder-decoders for separate recovery. The CNN features of each encoder are learned to capture either missing structures or textures without considering them as a whole. The insufficient utilization of these encoder features limit the performance of recovering both structures and textures. In this paper, we propose a mutual encoder-decoder CNN for joint recovery of both. We use CNN features from the deep and shallow layers of the encoder to represent structures and textures of an input image, respectively. The deep layer features are sent to a structure branch and the shallow layer features are sent to a texture branch. In each branch, we fill holes in multiple scales of the CNN features. The filled CNN features from both branches are concatenated and then equalized. During feature equalization, we reweigh channel attentions first and propose a bilateral propagation activation function to enable spatial equalization. To this end, the filled CNN features of structure and texture mutually benefit each other to represent image content at all feature levels. We use the equalized feature to supplement decoder features for output image generation through skip connections. Experiments on the benchmark datasets show the proposed method is effective to recover structures and textures and performs favorably against state-of-the-art approaches.

Hongyu Liu, Bin Jiang, Wei Huang et al.

Recent deep learning based image inpainting methods which utilize contextual information and two-stage architecture have exhibited remarkable performance. However, the two-stage architecture is time-consuming, the contextual information lack high-level semantics and ignores both the semantic relevance and distance information of hole's feature patches, these limitations result in blurry textures and distorted structures of final result. Motivated by these observations, we propose a new deep generative model-based approach, which trains a shared network twice with different targets and utilizes a single network during the testing phase, so that we can effectively save inference time. Specifically, the targets of two training steps are structure reconstruction and texture generation respectively. During the second training, we first propose a Pyramid Filling Block (PF-block) to utilize the high-level features that the hole regions has been filled to guide the filling process of low-level features progressively, the missing content can be filled from deep to shallow in a pyramid fashion. Then, inspired by the classical bilateral filter [30], we propose the Bilateral Attention layer (BA-layer) to optimize filled feature map, which synthesizes feature patches at each position by computing weighted sums of the surrounding feature patches, these weights are derived by considering both distance and value relationships between feature patches, thus making the visually plausible inpainting results. Finally, experiments on multiple publicly available datasets show the superior performance of our approach.

W. Ronny Huang, Zeyad Emam, Micah Goldblum et al.

The power of neural networks lies in their ability to generalize to unseen data, yet the underlying reasons for this phenomenon remain elusive. Numerous rigorous attempts have been made to explain generalization, but available bounds are still quite loose, and analysis does not always lead to true understanding. The goal of this work is to make generalization more intuitive. Using visualization methods, we discuss the mystery of generalization, the geometry of loss landscapes, and how the curse (or, rather, the blessing) of dimensionality causes optimizers to settle into minima that generalize well.

Karthik A. Sankararaman, Soham De, Zheng Xu et al.

This paper studies how neural network architecture affects the speed of training. We introduce a simple concept called gradient confusion to help formally analyze this. When gradient confusion is high, stochastic gradients produced by different data samples may be negatively correlated, slowing down convergence. But when gradient confusion is low, data samples interact harmoniously, and training proceeds quickly. Through theoretical and experimental results, we demonstrate how the neural network architecture affects gradient confusion, and thus the efficiency of training. Our results show that, for popular initialization techniques, increasing the width of neural networks leads to lower gradient confusion, and thus faster model training. On the other hand, increasing the depth of neural networks has the opposite effect. Our results indicate that alternate initialization techniques or networks using both batch normalization and skip connections help reduce the training burden of very deep networks.