Search

Diffusion Transformers (DiTs) have exhibited robust capabilities in image generation tasks. However, accurate text-guided image editing for multimodal DiTs (MM-DiTs) still poses a significant challenge. Unlike UNet-based structures that could utilize self/cross-attention maps for semantic editing, MM-DiTs inherently lack support for explicit and consistent incorporated text guidance, resulting in semantic misalignment between the edited results and texts. In this study, we disclose the sensitivity of different attention heads to different image semantics within MM-DiTs and introduce HeadRouter, a training-free image editing framework that edits the source image by adaptively routing the text guidance to different attention heads in MM-DiTs. Furthermore, we present a dual-token refinement module to refine text/image token representations for precise semantic guidance and accurate region expression. Experimental results on multiple benchmarks demonstrate HeadRouter's performance in terms of editing fidelity and image quality.

When training neural networks with custom objectives, such as ranking losses and shortest-path losses, a common problem is that they are, per se, non-differentiable. A popular approach is to continuously relax the objectives to provide gradients, enabling learning. However, such differentiable relaxations are often non-convex and can exhibit vanishing and exploding gradients, making them (already in isolation) hard to optimize. Here, the loss function poses the bottleneck when training a deep neural network. We present Newton Losses, a method for improving the performance of existing hard to optimize losses by exploiting their second-order information via their empirical Fisher and Hessian matrices. Instead of training the neural network with second-order techniques, we only utilize the loss function's second-order information to replace it by a Newton Loss, while training the network with gradient descent. This makes our method computationally efficient. We apply Newton Losses to eight differentiable algorithms for sorting and shortest-paths, achieving significant improvements for less-optimized differentiable algorithms, and consistent improvements, even for well-optimized differentiable algorithms., Comment: Published at NeurIPS 2024

A sequential solver for differential-algebraic equations with embedded optimization criteria (DAEOs) was developed to take advantage of the theoretical work done by Deussen et al. Solvers of this type separate the optimization problem from the differential equation and solve each individually. The new solver relies on the reduction of a DAEO to a sequence of differential inclusions separated by jump events. These jump events occur when the global solution to the optimization problem jumps to a new value. Without explicit treatment, these events will reduce the order of convergence of the integration step to one. The solver implements a "local optimizer tracking" procedure to detect and correct these jump events. Local optimizer tracking is much less expensive than running a deterministic global optimizer at every time step. This preserves the order of convergence of the integrator component without sacrificing performance to perform deterministic global optimization at every time step. The newly developed solver produces correct solutions to DAEOs and runs much faster than sequential DAEO solvers that rely only on global optimization., Comment: 8 pages, 5 figures

Current research provides methods to communicate uncertainty and adapts classical algorithms of the visualization pipeline to take the uncertainty into account. Various existing visualization frameworks include methods to present uncertain data but do not offer transformation techniques tailored to uncertain data. Therefore, we propose a software package for uncertainty-aware data analysis in Python (UADAPy) offering methods for uncertain data along the visualization pipeline. We aim to provide a platform that is the foundation for further integration of uncertainty algorithms and visualizations. It provides common utility functionality to support research in uncertainty-aware visualization algorithms and makes state-of-the-art research results accessible to the end user. The project is available at https://github.com/UniStuttgart-VISUS/uadapy.

Categorical data does not have an intrinsic definition of distance or order, and therefore, established visualization techniques for categorical data only allow for a set-based or frequency-based analysis, e.g., through Euler diagrams or Parallel Sets, and do not support a similarity-based analysis. We present a novel dimensionality reduction-based visualization for categorical data, which is based on defining the distance of two data items as the number of varying attributes. Our technique enables users to pre-attentively detect groups of similar data items and observe the properties of the projection, such as attributes strongly influencing the embedding. Our prototype visually encodes data properties in an enhanced scatterplot-like visualization, encoding attributes in the background to show the distribution of categories. In addition, we propose two graph-based measures to quantify the plot's visual quality, which rank attributes according to their contribution to cluster cohesion. To demonstrate the capabilities of our similarity-based approach, we compare it to Euler diagrams and Parallel Sets regarding visual scalability and show its benefits through an expert study with five data scientists analyzing the Titanic and Mushroom datasets with up to 23 attributes and 8124 category combinations. Our results indicate that the Categorical Data Map offers an effective analysis method, especially for large datasets with a high number of category combinations., Comment: Fully replaced; 10 pages, 9 figures, LaTeX; to appear at Visual Data Science (VDS) Symposium at IEEE VIS 2024

Personalized generation paradigms empower designers to customize visual intellectual properties with the help of textual descriptions by tuning or adapting pre-trained text-to-image models on a few images. Recent works explore approaches for concurrently customizing both content and detailed visual style appearance. However, these existing approaches often generate images where the content and style are entangled. In this study, we reconsider the customization of content and style concepts from the perspective of parameter space construction. Unlike existing methods that utilize a shared parameter space for content and style, we propose a learning framework that separates the parameter space to facilitate individual learning of content and style, thereby enabling disentangled content and style. To achieve this goal, we introduce "partly learnable projection" (PLP) matrices to separate the original adapters into divided sub-parameter spaces. We propose "break-for-make" customization learning pipeline based on PLP, which is simple yet effective. We break the original adapters into "up projection" and "down projection", train content and style PLPs individually with the guidance of corresponding textual prompts in the separate adapters, and maintain generalization by employing a multi-correspondence projection learning strategy. Based on the adapters broken apart for separate training content and style, we then make the entity parameter space by reconstructing the content and style PLPs matrices, followed by fine-tuning the combined adapter to generate the target object with the desired appearance. Experiments on various styles, including textures, materials, and artistic style, show that our method outperforms state-of-the-art single/multiple concept learning pipelines in terms of content-style-prompt alignment.

Large language models (LLMs) are widely deployed in various downstream tasks, e.g., auto-completion, aided writing, or chat-based text generation. However, the considered output candidates of the underlying search algorithm are under-explored and under-explained. We tackle this shortcoming by proposing a tree-in-the-loop approach, where a visual representation of the beam search tree is the central component for analyzing, explaining, and adapting the generated outputs. To support these tasks, we present generAItor, a visual analytics technique, augmenting the central beam search tree with various task-specific widgets, providing targeted visualizations and interaction possibilities. Our approach allows interactions on multiple levels and offers an iterative pipeline that encompasses generating, exploring, and comparing output candidates, as well as fine-tuning the model based on adapted data. Our case study shows that our tool generates new insights in gender bias analysis beyond state-of-the-art template-based methods. Additionally, we demonstrate the applicability of our approach in a qualitative user study. Finally, we quantitatively evaluate the adaptability of the model to few samples, as occurring in text-generation use cases., Comment: 24 pages paper, 4 pages references, 3 pages appendix, 8 figures

We propose a new approach for propagating stable probability distributions through neural networks. Our method is based on local linearization, which we show to be an optimal approximation in terms of total variation distance for the ReLU non-linearity. This allows propagating Gaussian and Cauchy input uncertainties through neural networks to quantify their output uncertainties. To demonstrate the utility of propagating distributions, we apply the proposed method to predicting calibrated confidence intervals and selective prediction on out-of-distribution data. The results demonstrate a broad applicability of propagating distributions and show the advantages of our method over other approaches such as moment matching., Comment: Published at ICLR 2024, Code @ https://github.com/Felix-Petersen/distprop

The seamless integration of music with dance movements is essential for communicating the artistic intent of a dance piece. This alignment also significantly improves the immersive quality of gaming experiences and animation productions. Although there has been remarkable advancement in creating high-fidelity music from textual descriptions, current methodologies mainly focus on modulating overall characteristics such as genre and emotional tone. They often overlook the nuanced management of temporal rhythm, which is indispensable in crafting music for dance, since it intricately aligns the musical beats with the dancers' movements. Recognizing this gap, we propose an encoder-based textual inversion technique to augment text-to-music models with visual control, facilitating personalized music generation. Specifically, we develop dual-path rhythm-genre inversion to effectively integrate the rhythm and genre of a dance motion sequence into the textual space of a text-to-music model. Contrary to traditional textual inversion methods, which directly update text embeddings to reconstruct a single target object, our approach utilizes separate rhythm and genre encoders to obtain text embeddings for two pseudo-words, adapting to the varying rhythms and genres. We collect a new dataset called In-the-wild Dance Videos (InDV) and demonstrate that our approach outperforms state-of-the-art methods across multiple evaluation metrics. Furthermore, our method is able to adapt to changes in tempo and effectively integrates with the inherent text-guided generation capability of the pre-trained model. Our source code and demo videos are available at \url{https://github.com/lsfhuihuiff/Dance-to-music_Siggraph_Asia_2024}, Comment: 11 pages, 5 figures, SIGGRAPH ASIA 2024

The growing popularity of generative language models has amplified interest in interactive methods to guide model outputs. Prompt refinement is considered one of the most effective means to influence output among these methods. We identify several challenges associated with prompting large language models, categorized into data- and model-specific, linguistic, and socio-linguistic challenges. A comprehensive examination of model outputs, including runner-up candidates and their corresponding probabilities, is needed to address these issues. The beam search tree, the prevalent algorithm to sample model outputs, can inherently supply this information. Consequently, we introduce an interactive visual method for investigating the beam search tree, facilitating analysis of the decisions made by the model during generation. We quantitatively show the value of exposing the beam search tree and present five detailed analysis scenarios addressing the identified challenges. Our methodology validates existing results and offers additional insights., Comment: 9 pages paper, 2 pages references, 7 figures

Markerless methods for animal posture tracking have been rapidly developing recently, but frameworks and benchmarks for tracking large animal groups in 3D are still lacking. To overcome this gap in the literature, we present 3D-MuPPET, a framework to estimate and track 3D poses of up to 10 pigeons at interactive speed using multiple camera views. We train a pose estimator to infer 2D keypoints and bounding boxes of multiple pigeons, then triangulate the keypoints to 3D. For identity matching of individuals in all views, we first dynamically match 2D detections to global identities in the first frame, then use a 2D tracker to maintain IDs across views in subsequent frames. We achieve comparable accuracy to a state of the art 3D pose estimator in terms of median error and Percentage of Correct Keypoints. Additionally, we benchmark the inference speed of 3D-MuPPET, with up to 9.45 fps in 2D and 1.89 fps in 3D, and perform quantitative tracking evaluation, which yields encouraging results. Finally, we showcase two novel applications for 3D-MuPPET. First, we train a model with data of single pigeons and achieve comparable results in 2D and 3D posture estimation for up to 5 pigeons. Second, we show that 3D-MuPPET also works in outdoors without additional annotations from natural environments. Both use cases simplify the domain shift to new species and environments, largely reducing annotation effort needed for 3D posture tracking. To the best of our knowledge we are the first to present a framework for 2D/3D animal posture and trajectory tracking that works in both indoor and outdoor environments for up to 10 individuals. We hope that the framework can open up new opportunities in studying animal collective behaviour and encourages further developments in 3D multi-animal posture tracking.

Line-based density plots are used to reduce visual clutter in line charts with a multitude of individual lines. However, these traditional density plots are often perceived ambiguously, which obstructs the user's identification of underlying trends in complex datasets. Thus, we propose a novel image space coloring method for line-based density plots that enhances their interpretability. Our method employs color not only to visually communicate data density but also to highlight similar regions in the plot, allowing users to identify and distinguish trends easily. We achieve this by performing hierarchical clustering based on the lines passing through each region and mapping the identified clusters to the hue circle using circular MDS. Additionally, we propose a heuristic approach to assign each line to the most probable cluster, enabling users to analyze density and individual lines. We motivate our method by conducting a small-scale user study, demonstrating the effectiveness of our method using synthetic and real-world datasets, and providing an interactive online tool for generating colored line-based density plots., Comment: Published in IEEE Transactions on Visualization and Computer Graphics (Supplementary Material: https://osf.io/jm5yz/)

Personalizing generative models offers a way to guide image generation with user-provided references. Current personalization methods can invert an object or concept into the textual conditioning space and compose new natural sentences for text-to-image diffusion models. However, representing and editing specific visual attributes such as material, style, and layout remains a challenge, leading to a lack of disentanglement and editability. To address this problem, we propose a novel approach that leverages the step-by-step generation process of diffusion models, which generate images from low to high frequency information, providing a new perspective on representing, generating, and editing images. We develop the Prompt Spectrum Space P*, an expanded textual conditioning space, and a new image representation method called \sysname. ProSpect represents an image as a collection of inverted textual token embeddings encoded from per-stage prompts, where each prompt corresponds to a specific generation stage (i.e., a group of consecutive steps) of the diffusion model. Experimental results demonstrate that P* and ProSpect offer better disentanglement and controllability compared to existing methods. We apply ProSpect in various personalized attribute-aware image generation applications, such as image-guided or text-driven manipulations of materials, style, and layout, achieving previously unattainable results from a single image input without fine-tuning the diffusion models. Our source code is available athttps://github.com/zyxElsa/ProSpect.

We are considering differential-algebraic equations with embedded optimization criteria (DAEOs) in which the embedded optimization problem is solved by global optimization. This actually leads to differential inclusions for cases in which there are multiple global optimizer at the same time. Jump events from one global optimum to another result in nonsmooth DAEs and thus reduction of the convergence order of the numerical integrator to first-order. Implementation of event detection and location as introduced in this work preserves the higher-order convergence behavior of the integrator. This allows to compute discrete tangents and adjoint sensitivities for optimal control problems., Comment: 16 pages, 4 figures

We present ISAAC (Input-baSed ApproximAte Curvature), a novel method that conditions the gradient using selected second-order information and has an asymptotically vanishing computational overhead, assuming a batch size smaller than the number of neurons. We show that it is possible to compute a good conditioner based on only the input to a respective layer without a substantial computational overhead. The proposed method allows effective training even in small-batch stochastic regimes, which makes it competitive to first-order as well as second-order methods., Comment: Published at ICLR 2023, Code @ https://github.com/Felix-Petersen/isaac, Video @ https://youtu.be/7RKRX-MdwqM

Visualization researchers and visualization professionals seek appropriate abstractions of visualization requirements that permit considering visualization solutions independently from specific problems. Abstractions can help us design, analyze, organize, and evaluate the things we create. The literature has many task structures (taxonomies, typologies, etc.), design spaces, and related ``frameworks'' that provide abstractions of the problems a visualization is meant to address. In this viewpoint, we introduce a different one, a problem space that complements existing frameworks by focusing on the needs that a visualization is meant to solve. We believe it provides a valuable conceptual tool for designing and discussing visualizations., Comment: Author's submitted version. An article with the same content was approved for publication by the Visualization Viewpoints Department of IEEE Computer Graphics and Applications magazine

In this paper, we propose the t-FDP model, a force-directed placement method based on a novel bounded short-range force (t-force) defined by Student's t-distribution. Our formulation is flexible, exerts limited repulsive forces for nearby nodes and can be adapted separately in its short- and long-range effects. Using such forces in force-directed graph layouts yields better neighborhood preservation than current methods, while maintaining low stress errors. Our efficient implementation using a Fast Fourier Transform is one order of magnitude faster than state-of-the-art methods and two orders faster on the GPU, enabling us to perform parameter tuning by globally and locally adjusting the t-force in real-time for complex graphs. We demonstrate the quality of our approach by numerical evaluation against state-of-the-art approaches and extensions for interactive exploration., Comment: To appear in IEEE Transactions on Visualization and Computer Graphics

Color is one of the main visual channels used for highlighting elements of interest in visualization. However, in multi-class scatterplots, color highlighting often comes at the expense of degraded color discriminability. In this paper, we argue for context-preserving highlighting during the interactive exploration of multi-class scatterplots to achieve desired pop-out effects, while maintaining good perceptual separability among all classes and consistent color mapping schemes under varying points of interest. We do this by first generating two contrastive color mapping schemes with large and small contrasts to the background. Both schemes maintain good perceptual separability among all classes and ensure that when colors from the two palettes are assigned to the same class, they have a high color consistency in color names. We then interactively combine these two schemes to create a dynamic color mapping for highlighting different points of interest. We demonstrate the effectiveness through crowd-sourced experiments and case studies., Comment: To appear in CHI'23: ACM Conference on Human Factors in Computing Systems

Recently, research has increasingly focused on developing efficient neural network architectures. In this work, we explore logic gate networks for machine learning tasks by learning combinations of logic gates. These networks comprise logic gates such as "AND" and "XOR", which allow for very fast execution. The difficulty in learning logic gate networks is that they are conventionally non-differentiable and therefore do not allow training with gradient descent. Thus, to allow for effective training, we propose differentiable logic gate networks, an architecture that combines real-valued logics and a continuously parameterized relaxation of the network. The resulting discretized logic gate networks achieve fast inference speeds, e.g., beyond a million images of MNIST per second on a single CPU core., Comment: Published at NeurIPS 2022

The top-k classification accuracy is one of the core metrics in machine learning. Here, k is conventionally a positive integer, such as 1 or 5, leading to top-1 or top-5 training objectives. In this work, we relax this assumption and optimize the model for multiple k simultaneously instead of using a single k. Leveraging recent advances in differentiable sorting and ranking, we propose a differentiable top-k cross-entropy classification loss. This allows training the network while not only considering the top-1 prediction, but also, e.g., the top-2 and top-5 predictions. We evaluate the proposed loss function for fine-tuning on state-of-the-art architectures, as well as for training from scratch. We find that relaxing k does not only produce better top-5 accuracies, but also leads to top-1 accuracy improvements. When fine-tuning publicly available ImageNet models, we achieve a new state-of-the-art for these models., Comment: Published at ICML 2022, Code @ https://github.com/Felix-Petersen/difftopk

In this work, we present and study a generalized family of differentiable renderers. We discuss from scratch which components are necessary for differentiable rendering and formalize the requirements for each component. We instantiate our general differentiable renderer, which generalizes existing differentiable renderers like SoftRas and DIB-R, with an array of different smoothing distributions to cover a large spectrum of reasonable settings. We evaluate an array of differentiable renderer instantiations on the popular ShapeNet 3D reconstruction benchmark and analyze the implications of our results. Surprisingly, the simple uniform distribution yields the best overall results when averaged over 13 classes; in general, however, the optimal choice of distribution heavily depends on the task., Comment: Published at CVPR 2022, Code @ https://github.com/Felix-Petersen/gendr

Differentiable sorting algorithms allow training with sorting and ranking supervision, where only the ordering or ranking of samples is known. Various methods have been proposed to address this challenge, ranging from optimal transport-based differentiable Sinkhorn sorting algorithms to making classic sorting networks differentiable. One problem of current differentiable sorting methods is that they are non-monotonic. To address this issue, we propose a novel relaxation of conditional swap operations that guarantees monotonicity in differentiable sorting networks. We introduce a family of sigmoid functions and prove that they produce differentiable sorting networks that are monotonic. Monotonicity ensures that the gradients always have the correct sign, which is an advantage in gradient-based optimization. We demonstrate that monotonic differentiable sorting networks improve upon previous differentiable sorting methods., Comment: Published at ICLR 2022, Code @ https://github.com/Felix-Petersen/diffsort, Video @ https://www.youtube.com/watch?v=Rl-sFaE1z4M

Reconstructing the 3D geometry of an object from an image is a major challenge in computer vision. Recently introduced differentiable renderers can be leveraged to learn the 3D geometry of objects from 2D images, but those approaches require additional supervision to enable the renderer to produce an output that can be compared to the input image. This can be scene information or constraints such as object silhouettes, uniform backgrounds, material, texture, and lighting. In this paper, we propose an approach that enables a differentiable rendering-based learning of 3D objects from images with backgrounds without the need for silhouette supervision. Instead of trying to render an image close to the input, we propose an adversarial style-transfer and domain adaptation pipeline that allows to translate the input image domain to the rendered image domain. This allows us to directly compare between a translated image and the differentiable rendering of a 3D object reconstruction in order to train the 3D object reconstruction network. We show that the approach learns 3D geometry from images with backgrounds and provides a better performance than constrained methods for single-view 3D object reconstruction on this task., Comment: To be published at WACV 2022, Code @ https://github.com/Felix-Petersen/style-agnostic-3d-reconstruction

The integration of algorithmic components into neural architectures has gained increased attention recently, as it allows training neural networks with new forms of supervision such as ordering constraints or silhouettes instead of using ground truth labels. Many approaches in the field focus on the continuous relaxation of a specific task and show promising results in this context. But the focus on single tasks also limits the applicability of the proposed concepts to a narrow range of applications. In this work, we build on those ideas to propose an approach that allows to integrate algorithms into end-to-end trainable neural network architectures based on a general approximation of discrete conditions. To this end, we relax these conditions in control structures such as conditional statements, loops, and indexing, so that resulting algorithms are smoothly differentiable. To obtain meaningful gradients, each relevant variable is perturbed via logistic distributions and the expectation value under this perturbation is approximated. We evaluate the proposed continuous relaxation model on four challenging tasks and show that it can keep up with relaxations specifically designed for each individual task., Comment: Published at NeurIPS 2021, Code @ https://github.com/Felix-Petersen/algovision, Video @ https://www.youtube.com/watch?v=01ENzpkjOCE