Recent image degradation estimation methods have enabled single-image super-resolution (SR) approaches to better upsample real-world images. Among these methods, explicit kernel estimation approaches have demonstrated unprecedented performance at handling unknown degradations. Nonetheless, a number of limitations constrain their efficacy when used by downstream SR models. Specifically, this family of methods yields i) excessive inference time due to long per-image adaptation times and ii) inferior image fidelity due to kernel mismatch. In this work, we introduce a learning-to-learn approach that meta-learns from the information contained in a distribution of images, thereby enabling significantly faster adaptation to new images with substantially improved performance in both kernel estimation and image fidelity. Specifically, we meta-train a kernel-generating GAN, named MetaKernelGAN, on a range of tasks, such that when a new image is presented, the generator starts from an informed kernel estimate and the discriminator starts with a strong capability to distinguish between patch distributions. Compared with state-of-the-art methods, our experiments show that MetaKernelGAN better estimates the magnitude and covariance of the kernel, leading to state-of-the-art blind SR results within a similar computational regime when combined with a non-blind SR model. Through supervised learning of an unsupervised learner, our method maintains the generalizability of the unsupervised learner, improves the optimization stability of kernel estimation, and hence image adaptation, and leads to a faster inference with a speedup between 14.24 to 102.1x over existing methods.

Learning causal structure from observational data often assumes that we observe independent and identically distributed (i.\,i.\,d) data. The traditional approach aims to find a graphical representation that encodes the same set of conditional independence relationships as those present in the observed distribution. It is known that under i.\,i.\,d assumption, even with infinite data, there is a limit to how fine-grained a causal structure we can identify. To overcome this limitation, recent work has explored using data originating from different, related environments to learn richer causal structure. These approaches implicitly rely on the independent causal mechanisms (ICM) principle, which postulates that the mechanism giving rise to an effect given its causes and the mechanism which generates the causes do not inform or influence each other. Thus, components of the causal model can independently change from environment to environment. Despite its wide application in machine learning and causal inference, there is a lack of statistical formalization of the ICM principle and how it enables identification of richer causal structures from grouped data. Here we present new causal de Finetti theorems which offer a first statistical formalization of ICM principle and show how causal structure identification is possible from exchangeable data. Our work provides theoretical justification for a broad range of techniques leveraging multi-environment data to learn causal structure.

在梯度下降中，改变我们参数化的方式如何导致巨大的优化轨迹，从而引起令人惊讶的有意义的感应偏差：识别稀疏分类器或重建低级矩阵而无明确正规化。这种隐式正规化已经假设是深入学习良好概括的贡献因素。然而，自然梯度下降近似不变于Reparameterization，它始终遵循相同的轨迹并找到相同的最佳值。自然出现的问题：如果我们消除了参数化的角色，会发生什么，将找到哪个解决方案，发生了哪些新的属性？我们在逻辑损失和深层矩阵分解下，对深层线性网络进行自然梯度流动的行为。我们的一些发现扩展到非线性神经网络，具有足够但有限的参数化。我们证明存在学习问题，其中自然梯度下降失败概括，而具有正确架构的梯度下降则表现良好。

可靠地预测围绕自动赛车的参赛者车辆的动议对于有效和表现计划至关重要。尽管高度表现力，但深度神经网络是黑盒模型，使其在安全至关重要的应用（例如自动驾驶）中具有挑战性。在本文中，我们介绍了一种结构化的方式，以预测具有深神网络的对立赛车的运动。最终可能的输出轨迹集受到限制。因此，可以给出有关预测的质量保证。我们通过将模型与基于LSTM的编码器架构一起评估模型来报告该模型的性能，这些架构是从高保真硬件中获取的数据中获得的。拟议的方法的表现优于预测准确性的基线，但仍能履行质量保证。因此，该模型的强大现实应用已被证明。介绍的模型被部署在慕尼黑技术大学的Indy Automous Challenge 2021中。本研究中使用的代码可作为开放源软件提供，网址为www.github.com/tumftm/mixnet。

r \'{e} NYI两个分布之间的跨熵度量，即香农跨透明拷贝的概括，最近用作改进的深度学习生成对抗网络设计的损失函数。在这项工作中，我们检查了该度量的属性，并在固定分布之一以及两个分布属于指数族时得出封闭形式的表达式。我们还通过分析确定了固定高斯过程和有限的阿尔如本字母马尔可夫来源的跨凝结速率的公式。