变形金刚是一种深入学习语言模型，用于数据中心中的自然语言处理（NLP）服务。在变压器模型中，生成的预训练的变压器（GPT）在文本生成或自然语言生成（NLG）中取得了显着的性能，它需要在摘要阶段处理大型输入上下文，然后是产生一个生成阶段的一次单词。常规平台（例如GPU）专门用于在摘要阶段平行处理大型输入，但是由于其顺序特征，它们的性能在生成阶段显着降低。因此，需要一个有效的硬件平台来解决由文本生成的顺序特征引起的高潜伏期。在本文中，我们提出了DFX，这是一种多FPGA加速器，该设备在摘要和发电阶段中执行GPT-2模型端到端，并具有低延迟和高吞吐量。 DFX使用模型并行性和优化的数据流，这是模型和硬件感知的设备之间快速同时执行执行。其计算核心根据自定义说明运行，并提供GPT-2操作端到端。我们在四个Xilinx Alveo U280 FPGAS上实现了建议的硬件体系结构，并利用了高带宽内存（HBM）的所有频道，以及用于高硬件效率的最大计算资源数量。 DFX在现代GPT-2模型上实现了四个NVIDIA V100 GPU的5.58倍加速度和3.99倍的能效。 DFX的成本效益比GPU设备更具成本效益，这表明它是云数据中心中文本生成工作负载的有前途解决方案。

为了执行无条件的视频生成，我们必须学习现实世界的分布。为了综合高质量视频，各种研究试图学习噪声和视频之间的映射函数，包括最近的努力来分离运动分配和外观分布。然而，以前的方法在离散的固定间隔时间内学习运动动态，这与物体体的运动的连续性相反。在本文中，我们提出了一种新颖的视频生成方法，了解运动和外观的单独分布，前者由神经颂歌建模，以学习自然运动动态。具体地，我们采用两级方法，其中第一阶段将噪声向量转换为任意帧速率的一系列关键点，并且第二级基于给定的关键点序列和外观噪声向量来合成视频。我们的模型不仅定量优于最近的视频生成基线，而且还演示了多功能功能，例如动态帧速率操纵和两个数据集之间的运动传输，从而打开新的门以不同的视频生成应用。

尽管基于深度学习的面部相关模型成功显着，但这些模型仍然仅限于真正人类面的领域。另一方面，由于缺乏组织良好的数据集，由于缺乏组织的数据集，动画面的域已经不太积极地研究。在本文中，我们通过可控的合成动画模型介绍了一个大规模动画CeleBfaces数据集（AnimeCeleb），以提高动画面域的研究。为了促进数据生成过程，我们基于开放式3D软件和开发的注释系统构建半自动管道。这导致构建大型动画面部数据集，包括具有丰富注释的多姿态和多样式动画面。实验表明，我们的数据集适用于各种动画相关的任务，如头部重新创建和着色。

The 3D-aware image synthesis focuses on conserving spatial consistency besides generating high-resolution images with fine details. Recently, Neural Radiance Field (NeRF) has been introduced for synthesizing novel views with low computational cost and superior performance. While several works investigate a generative NeRF and show remarkable achievement, they cannot handle conditional and continuous feature manipulation in the generation procedure. In this work, we introduce a novel model, called Class-Continuous Conditional Generative NeRF ($\text{C}^{3}$G-NeRF), which can synthesize conditionally manipulated photorealistic 3D-consistent images by projecting conditional features to the generator and the discriminator. The proposed $\text{C}^{3}$G-NeRF is evaluated with three image datasets, AFHQ, CelebA, and Cars. As a result, our model shows strong 3D-consistency with fine details and smooth interpolation in conditional feature manipulation. For instance, $\text{C}^{3}$G-NeRF exhibits a Fr\'echet Inception Distance (FID) of 7.64 in 3D-aware face image synthesis with a $\text{128}^{2}$ resolution. Additionally, we provide FIDs of generated 3D-aware images of each class of the datasets as it is possible to synthesize class-conditional images with $\text{C}^{3}$G-NeRF.

Cellular automata (CA) captivate researchers due to teh emergent, complex individualized behavior that simple global rules of interaction enact. Recent advances in the field have combined CA with convolutional neural networks to achieve self-regenerating images. This new branch of CA is called neural cellular automata [1]. The goal of this project is to use the idea of idea of neural cellular automata to grow prediction machines. We place many different convolutional neural networks in a grid. Each conv net cell outputs a prediction of what the next state will be, and minimizes predictive error. Cells received their neighbors' colors and fitnesses as input. Each cell's fitness score described how accurate its predictions were. Cells could also move to explore their environment and some stochasticity was applied to movement.

There is a dramatic shortage of skilled labor for modern vineyards. The Vinum project is developing a mobile robotic solution to autonomously navigate through vineyards for winter grapevine pruning. This necessitates an autonomous navigation stack for the robot pruning a vineyard. The Vinum project is using the quadruped robot HyQReal. This paper introduces an architecture for a quadruped robot to autonomously move through a vineyard by identifying and approaching grapevines for pruning. The higher level control is a state machine switching between searching for destination positions, autonomously navigating towards those locations, and stopping for the robot to complete a task. The destination points are determined by identifying grapevine trunks using instance segmentation from a Mask Region-Based Convolutional Neural Network (Mask-RCNN). These detections are sent through a filter to avoid redundancy and remove noisy detections. The combination of these features is the basis for the proposed architecture.

Feature selection helps reduce data acquisition costs in ML, but the standard approach is to train models with static feature subsets. Here, we consider the dynamic feature selection (DFS) problem where a model sequentially queries features based on the presently available information. DFS is often addressed with reinforcement learning (RL), but we explore a simpler approach of greedily selecting features based on their conditional mutual information. This method is theoretically appealing but requires oracle access to the data distribution, so we develop a learning approach based on amortized optimization. The proposed method is shown to recover the greedy policy when trained to optimality and outperforms numerous existing feature selection methods in our experiments, thus validating it as a simple but powerful approach for this problem.

In this paper, we learn a diffusion model to generate 3D data on a scene-scale. Specifically, our model crafts a 3D scene consisting of multiple objects, while recent diffusion research has focused on a single object. To realize our goal, we represent a scene with discrete class labels, i.e., categorical distribution, to assign multiple objects into semantic categories. Thus, we extend discrete diffusion models to learn scene-scale categorical distributions. In addition, we validate that a latent diffusion model can reduce computation costs for training and deploying. To the best of our knowledge, our work is the first to apply discrete and latent diffusion for 3D categorical data on a scene-scale. We further propose to perform semantic scene completion (SSC) by learning a conditional distribution using our diffusion model, where the condition is a partial observation in a sparse point cloud. In experiments, we empirically show that our diffusion models not only generate reasonable scenes, but also perform the scene completion task better than a discriminative model. Our code and models are available at https://github.com/zoomin-lee/scene-scale-diffusion

We introduce a new tool for stochastic convex optimization (SCO): a Reweighted Stochastic Query (ReSQue) estimator for the gradient of a function convolved with a (Gaussian) probability density. Combining ReSQue with recent advances in ball oracle acceleration [CJJJLST20, ACJJS21], we develop algorithms achieving state-of-the-art complexities for SCO in parallel and private settings. For a SCO objective constrained to the unit ball in $\mathbb{R}^d$, we obtain the following results (up to polylogarithmic factors). We give a parallel algorithm obtaining optimization error $\epsilon_{\text{opt}}$ with $d^{1/3}\epsilon_{\text{opt}}^{-2/3}$ gradient oracle query depth and $d^{1/3}\epsilon_{\text{opt}}^{-2/3} + \epsilon_{\text{opt}}^{-2}$ gradient queries in total, assuming access to a bounded-variance stochastic gradient estimator. For $\epsilon_{\text{opt}} \in [d^{-1}, d^{-1/4}]$, our algorithm matches the state-of-the-art oracle depth of [BJLLS19] while maintaining the optimal total work of stochastic gradient descent. We give an $(\epsilon_{\text{dp}}, \delta)$-differentially private algorithm which, given $n$ samples of Lipschitz loss functions, obtains near-optimal optimization error and makes $\min(n, n^2\epsilon_{\text{dp}}^2 d^{-1}) + \min(n^{4/3}\epsilon_{\text{dp}}^{1/3}, (nd)^{2/3}\epsilon_{\text{dp}}^{-1})$ queries to the gradients of these functions. In the regime $d \le n \epsilon_{\text{dp}}^{2}$, where privacy comes at no cost in terms of the optimal loss up to constants, our algorithm uses $n + (nd)^{2/3}\epsilon_{\text{dp}}^{-1}$ queries and improves recent advancements of [KLL21, AFKT21]. In the moderately low-dimensional setting $d \le \sqrt n \epsilon_{\text{dp}}^{3/2}$, our query complexity is near-linear.

We propose a new causal inference framework to learn causal effects from multiple, decentralized data sources in a federated setting. We introduce an adaptive transfer algorithm that learns the similarities among the data sources by utilizing Random Fourier Features to disentangle the loss function into multiple components, each of which is associated with a data source. The data sources may have different distributions; the causal effects are independently and systematically incorporated. The proposed method estimates the similarities among the sources through transfer coefficients, and hence requiring no prior information about the similarity measures. The heterogeneous causal effects can be estimated with no sharing of the raw training data among the sources, thus minimizing the risk of privacy leak. We also provide minimax lower bounds to assess the quality of the parameters learned from the disparate sources. The proposed method is empirically shown to outperform the baselines on decentralized data sources with dissimilar distributions.