可靠的概括是安全ML和AI的核心。但是，了解神经网络何时以及如何推广仍然是该领域最重要的未解决问题之一。在这项工作中，我们进行了一项广泛的实证研究（2200个模型，16个任务），以研究计算理论中的见解是否可以预测实践中神经网络概括的局限性。我们证明，根据Chomsky层次结构进行分组任务使我们能够预测某些架构是否能够推广到分布外输入。这包括负面结果，即使大量数据和训练时间也不会导致任何非平凡的概括，尽管模型具有足够的能力完美地适合培训数据。我们的结果表明，对于我们的任务子集，RNN和变形金刚无法概括非规范的任务，LSTMS可以解决常规和反语言任务，并且只有通过结构化内存（例如堆栈或存储器磁带）可以增强的网络可以成功地概括了无上下文和上下文敏感的任务。

诸如最大熵正则化之类的政策正则化方法被广泛用于增强学习以提高学习政策的鲁棒性。在本文中，我们展示了这种鲁棒性是如何通过对冲的奖励功能扰动而产生的，奖励功能是从想象中的对手设定的限制设置中选择的。使用凸双重性，我们表征了KL和Alpha-Divergence正则化的一组强大的对抗奖励扰动集，其中包括香农和Tsallis熵正则定期为特殊情况。重要的是，可以在此强大集合中给出概括保证。我们提供了有关最坏的奖励扰动的详细讨论，并提供了直观的经验示例，以说明这种稳健性及其与概括的关系。最后，我们讨论我们的分析如何补充并扩展对对抗奖励鲁棒性和路径一致性最佳条件的先前结果。

我们扩展了时间差异（TD）学习，以获得风险敏感的无模型加强学习算法。该扩展可以被视为Rescorla-Wagner规则的修改，其中（六样）刺激被认为是过度或低估TD目标的事件。结果，获得从I.I.D的自由能量的随机近似规则。通过高斯分布产生的样本，具有未知的平均值和方差。由于已知高斯自由能量是对平均值和方差的确定性相当敏感，因此学习规则具有风险敏感决策的应用。

In the last years, the number of IoT devices deployed has suffered an undoubted explosion, reaching the scale of billions. However, some new cybersecurity issues have appeared together with this development. Some of these issues are the deployment of unauthorized devices, malicious code modification, malware deployment, or vulnerability exploitation. This fact has motivated the requirement for new device identification mechanisms based on behavior monitoring. Besides, these solutions have recently leveraged Machine and Deep Learning techniques due to the advances in this field and the increase in processing capabilities. In contrast, attackers do not stay stalled and have developed adversarial attacks focused on context modification and ML/DL evaluation evasion applied to IoT device identification solutions. This work explores the performance of hardware behavior-based individual device identification, how it is affected by possible context- and ML/DL-focused attacks, and how its resilience can be improved using defense techniques. In this sense, it proposes an LSTM-CNN architecture based on hardware performance behavior for individual device identification. Then, previous techniques have been compared with the proposed architecture using a hardware performance dataset collected from 45 Raspberry Pi devices running identical software. The LSTM-CNN improves previous solutions achieving a +0.96 average F1-Score and 0.8 minimum TPR for all devices. Afterward, context- and ML/DL-focused adversarial attacks were applied against the previous model to test its robustness. A temperature-based context attack was not able to disrupt the identification. However, some ML/DL state-of-the-art evasion attacks were successful. Finally, adversarial training and model distillation defense techniques are selected to improve the model resilience to evasion attacks, without degrading its performance.

Cybercriminals are moving towards zero-day attacks affecting resource-constrained devices such as single-board computers (SBC). Assuming that perfect security is unrealistic, Moving Target Defense (MTD) is a promising approach to mitigate attacks by dynamically altering target attack surfaces. Still, selecting suitable MTD techniques for zero-day attacks is an open challenge. Reinforcement Learning (RL) could be an effective approach to optimize the MTD selection through trial and error, but the literature fails when i) evaluating the performance of RL and MTD solutions in real-world scenarios, ii) studying whether behavioral fingerprinting is suitable for representing SBC's states, and iii) calculating the consumption of resources in SBC. To improve these limitations, the work at hand proposes an online RL-based framework to learn the correct MTD mechanisms mitigating heterogeneous zero-day attacks in SBC. The framework considers behavioral fingerprinting to represent SBCs' states and RL to learn MTD techniques that mitigate each malicious state. It has been deployed on a real IoT crowdsensing scenario with a Raspberry Pi acting as a spectrum sensor. More in detail, the Raspberry Pi has been infected with different samples of command and control malware, rootkits, and ransomware to later select between four existing MTD techniques. A set of experiments demonstrated the suitability of the framework to learn proper MTD techniques mitigating all attacks (except a harmfulness rootkit) while consuming <1 MB of storage and utilizing <55% CPU and <80% RAM.

A universal kernel is constructed whose sections approximate any causal and time-invariant filter in the fading memory category with inputs and outputs in a finite-dimensional Euclidean space. This kernel is built using the reservoir functional associated with a state-space representation of the Volterra series expansion available for any analytic fading memory filter. It is hence called the Volterra reservoir kernel. Even though the state-space representation and the corresponding reservoir feature map are defined on an infinite-dimensional tensor algebra space, the kernel map is characterized by explicit recursions that are readily computable for specific data sets when employed in estimation problems using the representer theorem. We showcase the performance of the Volterra reservoir kernel in a popular data science application in relation to bitcoin price prediction.

Stress has a great effect on people's lives that can not be understated. While it can be good, since it helps humans to adapt to new and different situations, it can also be harmful when not dealt with properly, leading to chronic stress. The objective of this paper is developing a stress monitoring solution, that can be used in real life, while being able to tackle this challenge in a positive way. The SMILE data set was provided to team Anxolotl, and all it was needed was to develop a robust model. We developed a supervised learning model for classification in Python, presenting the final result of 64.1% in accuracy and a f1-score of 54.96%. The resulting solution stood the robustness test, presenting low variation between runs, which was a major point for it's possible integration in the Anxolotl app in the future.

The task of reconstructing 3D human motion has wideranging applications. The gold standard Motion capture (MoCap) systems are accurate but inaccessible to the general public due to their cost, hardware and space constraints. In contrast, monocular human mesh recovery (HMR) methods are much more accessible than MoCap as they take single-view videos as inputs. Replacing the multi-view Mo- Cap systems with a monocular HMR method would break the current barriers to collecting accurate 3D motion thus making exciting applications like motion analysis and motiondriven animation accessible to the general public. However, performance of existing HMR methods degrade when the video contains challenging and dynamic motion that is not in existing MoCap datasets used for training. This reduces its appeal as dynamic motion is frequently the target in 3D motion recovery in the aforementioned applications. Our study aims to bridge the gap between monocular HMR and multi-view MoCap systems by leveraging information shared across multiple video instances of the same action. We introduce the Neural Motion (NeMo) field. It is optimized to represent the underlying 3D motions across a set of videos of the same action. Empirically, we show that NeMo can recover 3D motion in sports using videos from the Penn Action dataset, where NeMo outperforms existing HMR methods in terms of 2D keypoint detection. To further validate NeMo using 3D metrics, we collected a small MoCap dataset mimicking actions in Penn Action,and show that NeMo achieves better 3D reconstruction compared to various baselines.

The field of Automatic Music Generation has seen significant progress thanks to the advent of Deep Learning. However, most of these results have been produced by unconditional models, which lack the ability to interact with their users, not allowing them to guide the generative process in meaningful and practical ways. Moreover, synthesizing music that remains coherent across longer timescales while still capturing the local aspects that make it sound ``realistic'' or ``human-like'' is still challenging. This is due to the large computational requirements needed to work with long sequences of data, and also to limitations imposed by the training schemes that are often employed. In this paper, we propose a generative model of symbolic music conditioned by data retrieved from human sentiment. The model is a Transformer-GAN trained with labels that correspond to different configurations of the valence and arousal dimensions that quantitatively represent human affective states. We try to tackle both of the problems above by employing an efficient linear version of Attention and using a Discriminator both as a tool to improve the overall quality of the generated music and its ability to follow the conditioning signals.

Reinforcement Learning is a powerful tool to model decision-making processes. However, it relies on an exploration-exploitation trade-off that remains an open challenge for many tasks. In this work, we study neighboring state-based, model-free exploration led by the intuition that, for an early-stage agent, considering actions derived from a bounded region of nearby states may lead to better actions when exploring. We propose two algorithms that choose exploratory actions based on a survey of nearby states, and find that one of our methods, ${\rho}$-explore, consistently outperforms the Double DQN baseline in an discrete environment by 49\% in terms of Eval Reward Return.