Transformer models have demonstrated exceptional performance across a wide range of applications. Though forming the foundation of Transformer models, the dot-product attention does not scale well to long-context data since its time requirement grows quadratically with context length. In this work, we propose Radar, a training-free approach that accelerates inference by dynamically searching for the most important context tokens. For any pre-trained Transformer, Radar can reduce the decoding time complexity without training or heuristically evicting tokens. Moreover, we provide theoretical justification for our approach, demonstrating that Radar can reliably identify the most important tokens with high probability. We conduct extensive comparisons with the previous methods on a wide range of tasks. The results demonstrate that Radar achieves the state-of-the-art performance across different architectures with reduced time complexity, offering a practical solution for efficient long-context processing of Transformers. The code is publicly available at https://github.com/BorealisAI/radar-decoding.

The paper presents a novel approach for vector-floorplan generation via a diffusion model, which denoises 2D coordinates of room/door corners with two inference objectives: 1) a single-step noise as the continuous quantity to precisely invert the continuous forward process; and 2) the final 2D coordinate as the discrete quantity to establish geometric incident relationships such as parallelism, orthogonality, and corner-sharing. Our task is graph-conditioned floorplan generation, a common workflow in floorplan design. We represent a floorplan as 1D polygonal loops, each of which corresponds to a room or a door. Our diffusion model employs a Transformer architecture at the core, which controls the attention masks based on the input graph-constraint and directly generates vector-graphics floorplans via a discrete and continuous denoising process. We have evaluated our approach on RPLAN dataset. The proposed approach makes significant improvements in all the metrics against the state-of-the-art with significant margins, while being capable of generating non-Manhattan structures and controlling the exact number of corners per room. A project website with supplementary video and document is here https://aminshabani.github.io/housediffusion.

This paper proposes a generative adversarial layout refinement network for automated floorplan generation. Our architecture is an integration of a graph-constrained relational GAN and a conditional GAN, where a previously generated layout becomes the next input constraint, enabling iterative refinement. A surprising discovery of our research is that a simple non-iterative training process, dubbed component-wise GT-conditioning, is effective in learning such a generator. The iterative generator further allows us to improve a metric of choice via meta-optimization techniques by controlling when to pass which input constraints during iterative refinement. Our qualitative and quantitative evaluation based on the three standard metrics demonstrate that the proposed system makes significant improvements over the current state-of-the-art, even competitive against the ground-truth floorplans, designed by professional architects. Code, model, and data are available at https://ennauata.github.io/houseganpp/page.html.

The convolutional neural network (CNN) works very well in many computer vision tasks including the face-related problems. However, in the case of age estimation and facial expression recognition (FER), the accuracy provided by the CNN is still not good enough to be used for the real-world problems. It seems that the CNN does not well find the subtle differences in thickness and amount of wrinkles on the face, which are the essential features for the age estimation and FER. Also, the face images in the real world have many variations due to the face rotation and illumination, where the CNN is not robust in finding the rotated objects when not every possible variation is in the training data. To alleviate these problems, we first propose to use the Gabor filter responses of faces as the input to the CNN, along with the original face image. This method enhances the wrinkles on the face so that the face-related features are found in the earlier stage of convolutional layers, and hence the overall performance is increased. We also adopt the idea of capsule network, which is shown to be robust to the rotation of objects and be able to capture the relationship of facial landmarks. We show that the performance of age estimation and FER are improved by using the capsule network than using the plain CNNs. Moreover, by using the Gabor responses as the input to the capsule network, the overall performances of face-related problems are increased compared to the recent CNN-based methods.