AttnMod: Attention-Based New Art Styles

Imagine a human artist looking at the generated photo of a diffusion model, and hoping to create a painting out of it. There could be some feature of the object in the photo that the artist wants to emphasize, some color to disperse, some silhouette to twist, or some part of the scene to be materialized. These intentions can be viewed as the modification of the cross attention from the text prompt onto UNet, during the desoising diffusion. This work presents AttnMod, to modify attention for creating new unpromptable art styles out of existing diffusion models. The style-creating behavior is studied across different setups.

Latent Painter

Latent diffusers revolutionized the generative AI and inspired creative art. When denoising the latent, the predicted original image at each step collectively animates the formation. However, the animation is limited by the denoising nature of the diffuser, and only renders a sharpening process. This work presents Latent Painter, which uses the latent as the canvas, and the diffuser predictions as the plan, to generate painting animation. Latent Painter also transits one generated image to another, which can happen between images from two different sets of checkpoints.

Channel Decomposition on Generative Networks

This work presents a method to decompose a layer of the generative networks into the painting actions. To behave like the human painter, these actions are driven by the cost simulating the hand movement, the paint color change, the stroke shape and the stroking style. To help planning, the Mask R-CNN is applied to detect the object areas and decide the painting order. The proposed painting system introduces a variety of extensions in artistic styles, based on the chosen parameters. Further experiments are performed to evaluate the channel penetration and the channel sensitivity on the strokes.

Topical Behavior Prediction from Massive Logs

In this paper, we study the topical behavior in a large scale. We use the network logs where each entry contains the entity ID, the timestamp, and the meta data about the activity. Both the temporal and the spatial relationships of the behavior are explored with the deep learning architectures combing the recurrent neural network (RNN) and the convolutional neural network (CNN). To make the behavioral data appropriate for the spatial learning in the CNN, we propose several reduction steps to form the topical metrics and to place them homogeneously like pixels in the images. The experimental result shows both temporal and spatial gains when compared against a multilayer perceptron (MLP) network. A new learning framework called the spatially connected convolutional networks (SCCN) is introduced to predict the topical metrics more efficiently.

Summarized Network Behavior Prediction

This work studies the entity-wise topical behavior from massive network logs. Both the temporal and the spatial relationships of the behavior are explored with the learning architectures combing the recurrent neural network (RNN) and the convolutional neural network (CNN). To make the behavioral data appropriate for the spatial learning in CNN, several reduction steps are taken to form the topical metrics and place them homogeneously like pixels in the images. The experimental result shows both the temporal- and the spatial- gains when compared to a multilayer perceptron (MLP) network. A new learning framework called spatially connected convolutional networks (SCCN) is introduced to more efficiently predict the behavior.

Large Scale Behavioral Analytics via Topical Interaction

We propose the split-diffuse (SD) algorithm that takes the output of an existing dimension reduction algorithm, and distributes the data points uniformly across the visualization space. The result, called the topic grids, is a set of grids on various topics which are generated from the free-form text content of any domain of interest. The topic grids efficiently utilizes the visualization space to provide visual summaries for massive data. Topical analysis, comparison and interaction can be performed on the topic grids in a more perceivable way.

Interacting with Massive Behavioral Data

In this short paper, we propose the split-diffuse (SD) algorithm that takes the output of an existing word embedding algorithm, and distributes the data points uniformly across the visualization space. The result improves the perceivability and the interactability by the human. We apply the SD algorithm to analyze the user behavior through access logs within the cyber security domain. The result, named the topic grids, is a set of grids on various topics generated from the logs. On the same set of grids, different behavioral metrics can be shown on different targets over different periods of time, to provide visualization and interaction to the human experts. Analysis, investigation, and other types of interaction can be performed on the topic grids more efficiently than on the output of existing dimension reduction methods. In addition to the cyber security domain, the topic grids can be further applied to other domains like e-commerce, credit card transaction, customer service to analyze the behavior in a large scale.

Topic Grids for Homogeneous Data Visualization

We propose the topic grids to detect anomaly and analyze the behavior based on the access log content. Content-based behavioral risk is quantified in the high dimensional space where the topics are generated from the log. The topics are being projected homogeneously into a space that is perception- and interaction-friendly to the human experts.