众所周知，在漫长的地平线和稀疏的奖励任务中，加强学习（RL）是困难的，需要大量的培训步骤。加快该过程的标准解决方案是利用额外的奖励信号，将其塑造以更好地指导学习过程。在语言条件的RL的背景下，语言输入的抽象和概括属性为更有效地塑造奖励的方式提供了机会。在本文中，我们利用这一想法并提出了一种自动奖励塑形方法，代理商从一般语言目标中提取辅助目标。这些辅助目标使用问题生成（QG）和问题答案（QA）系统：它们包括导致代理商尝试使用其自己的轨迹重建有关全球目标的部分信息的问题。当它成功时，它会获得与对答案的信心成正比的内在奖励。这激励代理生成轨迹，这些轨迹明确解释了一般语言目标的各个方面。我们的实验研究表明，这种方法不需要工程师干预来设计辅助目标，可以通过有效指导探索来提高样品效率。

多进球的增强学习最近吸引了大量的研究兴趣。通过允许在相关培训任务之间共享经验，只要在被考虑的目标空间中存在某些平滑度时，这种设置有利于测试时间的新任务的概括。但是，在州或目标空间不连续的环境（例如迷宫中的墙壁）中，由于缺乏专家知识的稀疏性，大多数目标都难以实现。这意味着必须发现一些艰苦的探索，必须发现一些目标课程，以通过使培训任务适应其当前功能来帮助代理商学习。我们以最新的自动课程学习技术为目标政策，我们提出了一种新颖的方法：Stein变化目标生成（SVGG），该方法通过利用一种学识渊博的模型来寻求在代理的近端开发区域中的新目标，以寻求新的目标它的能力和目标分布在勘探空间中以颗粒为模型。我们的方法依靠Stein变分梯度下降来动态吸引适当难度领域的目标采样分布。与最近最新的RL方法相比，我们证明了该方法的性能，即目标领域的成功覆盖范围。

我们为生成对抗网络（GAN）提出了一个新颖的理论框架。我们揭示了先前分析的基本缺陷，通过错误地对GANS的训练计划进行了错误的建模，该缺陷受到定义不定的鉴别梯度的约束。我们克服了这个问题，该问题阻碍了对GAN培训的原则研究，并考虑了歧视者的体系结构在我们的框架内解决它。为此，我们通过其神经切线核为歧视者提供了无限宽度神经网络的理论。我们表征了训练有素的判别器，以实现广泛的损失，并建立网络的一般可怜性属性。由此，我们获得了有关生成分布的融合的新见解，从而促进了我们对GANS训练动态的理解。我们通过基于我们的框架的分析工具包来证实这些结果，并揭示了与GAN实践一致的直觉。

Non-linear state-space models, also known as general hidden Markov models, are ubiquitous in statistical machine learning, being the most classical generative models for serial data and sequences in general. The particle-based, rapid incremental smoother PaRIS is a sequential Monte Carlo (SMC) technique allowing for efficient online approximation of expectations of additive functionals under the smoothing distribution in these models. Such expectations appear naturally in several learning contexts, such as likelihood estimation (MLE) and Markov score climbing (MSC). PARIS has linear computational complexity, limited memory requirements and comes with non-asymptotic bounds, convergence results and stability guarantees. Still, being based on self-normalised importance sampling, the PaRIS estimator is biased. Our first contribution is to design a novel additive smoothing algorithm, the Parisian particle Gibbs PPG sampler, which can be viewed as a PaRIS algorithm driven by conditional SMC moves, resulting in bias-reduced estimates of the targeted quantities. We substantiate the PPG algorithm with theoretical results, including new bounds on bias and variance as well as deviation inequalities. Our second contribution is to apply PPG in a learning framework, covering MLE and MSC as special examples. In this context, we establish, under standard assumptions, non-asymptotic bounds highlighting the value of bias reduction and the implicit Rao--Blackwellization of PPG. These are the first non-asymptotic results of this kind in this setting. We illustrate our theoretical results with numerical experiments supporting our claims.

Long-term non-prehensile planar manipulation is a challenging task for robot planning and feedback control. It is characterized by underactuation, hybrid control, and contact uncertainty. One main difficulty is to determine contact points and directions, which involves joint logic and geometrical reasoning in the modes of the dynamics model. To tackle this issue, we propose a demonstration-guided hierarchical optimization framework to achieve offline task and motion planning (TAMP). Our work extends the formulation of the dynamics model of the pusher-slider system to include separation mode with face switching cases, and solves a warm-started TAMP problem by exploiting human demonstrations. We show that our approach can cope well with the local minima problems currently present in the state-of-the-art solvers and determine a valid solution to the task. We validate our results in simulation and demonstrate its applicability on a pusher-slider system with real Franka Emika robot in the presence of external disturbances.

The number of international benchmarking competitions is steadily increasing in various fields of machine learning (ML) research and practice. So far, however, little is known about the common practice as well as bottlenecks faced by the community in tackling the research questions posed. To shed light on the status quo of algorithm development in the specific field of biomedical imaging analysis, we designed an international survey that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences (80 competitions in total). The survey covered participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics. A median of 72% challenge participants took part in the survey. According to our results, knowledge exchange was the primary incentive (70%) for participation, while the reception of prize money played only a minor role (16%). While a median of 80 working hours was spent on method development, a large portion of participants stated that they did not have enough time for method development (32%). 25% perceived the infrastructure to be a bottleneck. Overall, 94% of all solutions were deep learning-based. Of these, 84% were based on standard architectures. 43% of the respondents reported that the data samples (e.g., images) were too large to be processed at once. This was most commonly addressed by patch-based training (69%), downsampling (37%), and solving 3D analysis tasks as a series of 2D tasks. K-fold cross-validation on the training set was performed by only 37% of the participants and only 50% of the participants performed ensembling based on multiple identical models (61%) or heterogeneous models (39%). 48% of the respondents applied postprocessing steps.

Many problems in robotics are fundamentally problems of geometry, which lead to an increased research effort in geometric methods for robotics in recent years. The results were algorithms using the various frameworks of screw theory, Lie algebra and dual quaternions. A unification and generalization of these popular formalisms can be found in geometric algebra. The aim of this paper is to showcase the capabilities of geometric algebra when applied to robot manipulation tasks. In particular the modelling of cost functions for optimal control can be done uniformly across different geometric primitives leading to a low symbolic complexity of the resulting expressions and a geometric intuitiveness. We demonstrate the usefulness, simplicity and computational efficiency of geometric algebra in several experiments using a Franka Emika robot. The presented algorithms were implemented in c++20 and resulted in the publicly available library \textit{gafro}. The benchmark shows faster computation of the kinematics than state-of-the-art robotics libraries.

In the upcoming years, artificial intelligence (AI) is going to transform the practice of medicine in most of its specialties. Deep learning can help achieve better and earlier problem detection, while reducing errors on diagnosis. By feeding a deep neural network (DNN) with the data from a low-cost and low-accuracy sensor array, we demonstrate that it becomes possible to significantly improve the measurements' precision and accuracy. The data collection is done with an array composed of 32 temperature sensors, including 16 analog and 16 digital sensors. All sensors have accuracies between 0.5-2.0$^\circ$C. 800 vectors are extracted, covering a range from to 30 to 45$^\circ$C. In order to improve the temperature readings, we use machine learning to perform a linear regression analysis through a DNN. In an attempt to minimize the model's complexity in order to eventually run inferences locally, the network with the best results involves only three layers using the hyperbolic tangent activation function and the Adam Stochastic Gradient Descent (SGD) optimizer. The model is trained with a randomly-selected dataset using 640 vectors (80% of the data) and tested with 160 vectors (20%). Using the mean squared error as a loss function between the data and the model's prediction, we achieve a loss of only 1.47x10$^{-4}$ on the training set and 1.22x10$^{-4}$ on the test set. As such, we believe this appealing approach offers a new pathway towards significantly better datasets using readily-available ultra low-cost sensors.

Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License.

Progress in machine learning (ML) comes with a cost to the environment, given that training ML models requires significant computational resources, energy and materials. In the present article, we aim to quantify the carbon footprint of BLOOM, a 176-billion parameter language model, across its life cycle. We estimate that BLOOM's final training emitted approximately 24.7 tonnes of~\carboneq~if we consider only the dynamic power consumption, and 50.5 tonnes if we account for all processes ranging from equipment manufacturing to energy-based operational consumption. We also study the energy requirements and carbon emissions of its deployment for inference via an API endpoint receiving user queries in real-time. We conclude with a discussion regarding the difficulty of precisely estimating the carbon footprint of ML models and future research directions that can contribute towards improving carbon emissions reporting.