We commonly assume that data are a homogeneous set of observations when learning the structure of Bayesian networks. However, they often comprise different data sets that are related but not homogeneous because they have been collected in different ways or from different populations. In our previous work (Azzimonti, Corani and Scutari, 2021), we proposed a closed-form Bayesian Hierarchical Dirichlet score for discrete data that pools information across related data sets to learn a single encompassing network structure, while taking into account the differences in their probabilistic structures. In this paper, we provide an analogous solution for learning a Bayesian network from continuous data using mixed-effects models to pool information across the related data sets. We study its structural, parametric, predictive and classification accuracy and we show that it outperforms both conditional Gaussian Bayesian networks (that do not perform any pooling) and classical Gaussian Bayesian networks (that disregard the heterogeneous nature of the data). The improvement is marked for low sample sizes and for unbalanced data sets.

Reinforcement learning (RL) has been applied to attack graphs for penetration testing, however, trained agents do not reflect reality because the attack graphs lack operational nuances typically captured within the intelligence preparation of the battlefield (IPB) that include notions of (cyber) terrain. In particular, current practice constructs attack graphs exclusively using the Common Vulnerability Scoring System (CVSS) and its components. We present methods for constructing attack graphs using notions from IPB on cyber terrain analysis of obstacles, avenues of approach, key terrain, observation and fields of fire, and cover and concealment. We demonstrate our methods on an example where firewalls are treated as obstacles and represented in (1) the reward space and (2) the state dynamics. We show that terrain analysis can be used to bring realism to attack graphs for RL.

To facilitate the development of new models to bridge the gap between machine and human social intelligence, the recently proposed Baby Intuitions Benchmark (arXiv:2102.11938) provides a suite of tasks designed to evaluate commonsense reasoning about agents' goals and actions that even young infants exhibit. Here we present a principled Bayesian solution to this benchmark, based on a hierarchically Bayesian Theory of Mind (HBToM). By including hierarchical priors on agent goals and dispositions, inference over our HBToM model enables few-shot learning of the efficiency and preferences of an agent, which can then be used in commonsense plausibility judgements about subsequent agent behavior. This approach achieves near-perfect accuracy on most benchmark tasks, outperforming deep learning and imitation learning baselines while producing interpretable human-like inferences, demonstrating the advantages of structured Bayesian models of human social cognition.

Traditional model-based reinforcement learning (RL) methods generate forward rollout traces using the learnt dynamics model to reduce interactions with the real environment. The recent model-based RL method considers the way to learn a backward model that specifies the conditional probability of the previous state given the previous action and the current state to additionally generate backward rollout trajectories. However, in this type of model-based method, the samples derived from backward rollouts and those from forward rollouts are simply aggregated together to optimize the policy via the model-free RL algorithm, which may decrease both the sample efficiency and the convergence rate. This is because such an approach ignores the fact that backward rollout traces are often generated starting from some high-value states and are certainly more instructive for the agent to improve the behavior. In this paper, we propose the backward imitation and forward reinforcement learning (BIFRL) framework where the agent treats backward rollout traces as expert demonstrations for the imitation of excellent behaviors, and then collects forward rollout transitions for policy reinforcement. Consequently, BIFRL empowers the agent to both reach to and explore from high-value states in a more efficient manner, and further reduces the real interactions, making it potentially more suitable for real-robot learning. Moreover, a value-regularized generative adversarial network is introduced to augment the valuable states which are infrequently received by the agent. Theoretically, we provide the condition where BIFRL is superior to the baseline methods. Experimentally, we demonstrate that BIFRL acquires the better sample efficiency and produces the competitive asymptotic performance on various MuJoCo locomotion tasks compared against state-of-the-art model-based methods.

Cardiac imaging known as echocardiography is a non-invasive tool utilized to produce data including images and videos, which cardiologists use to diagnose cardiac abnormalities in general and myocardial infarction (MI) in particular. Echocardiography machines can deliver abundant amounts of data that need to be quickly analyzed by cardiologists to help them make a diagnosis and treat cardiac conditions. However, the acquired data quality varies depending on the acquisition conditions and the patient's responsiveness to the setup instructions. These constraints are challenging to doctors especially when patients are facing MI and their lives are at stake. In this paper, we propose an innovative real-time end-to-end fully automated model based on convolutional neural networks (CNN) to detect MI depending on regional wall motion abnormalities (RWMA) of the left ventricle (LV) from videos produced by echocardiography. Our model is implemented as a pipeline consisting of a 2D CNN that performs data preprocessing by segmenting the LV chamber from the apical four-chamber (A4C) view, followed by a 3D CNN that performs a binary classification to detect if the segmented echocardiography shows signs of MI. We trained both CNNs on a dataset composed of 165 echocardiography videos each acquired from a distinct patient. The 2D CNN achieved an accuracy of 97.18% on data segmentation while the 3D CNN achieved 90.9% of accuracy, 100% of precision and 95% of recall on MI detection. Our results demonstrate that creating a fully automated system for MI detection is feasible and propitious.

Strong AI requires the learning engine to be task non-specific and to automatically construct a dynamic hierarchy of internal features. By hierarchy, we mean, e.g., short road edges and short bush edges amount to intermediate features of landmarks; but intermediate features from tree shadows are distractors that must be disregarded by the high-level landmark concept. By dynamic, we mean the automatic selection of features while disregarding distractors is not static, but instead based on dynamic statistics (e.g. because of the instability of shadows in the context of landmark). By internal features, we mean that they are not only sensory, but also motor, so that context from motor (state) integrates with sensory inputs to become a context-based logic machine. We present why strong AI is necessary for any practical AI systems that work reliably in the real world. We then present a new generation of Developmental Networks 2 (DN-2). With many new novelties beyond DN-1, the most important novelty of DN-2 is that the inhibition area of each internal neuron is neuron-specific and dynamic. This enables DN-2 to automatically construct an internal hierarchy that is fluid, whose number of areas is not static as in DN-1. To optimally use the limited resource available, we establish that DN-2 is optimal in terms of maximum likelihood, under the condition of limited learning experience and limited resources. We also present how DN-2 can learn an emergent Universal Turing Machine (UTM). Together with the optimality, we present the optimal UTM. Experiments for real-world vision-based navigation, maze planning, and audition used DN-2. They successfully showed that DN-2 is for general purposes using natural and synthetic inputs. Their automatically constructed internal representation focuses on important features while being invariant to distractors and other irrelevant context-concepts.

With the development of hardware and algorithms, ASR(Automatic Speech Recognition) systems evolve a lot. As The models get simpler, the difficulty of development and deployment become easier, ASR systems are getting closer to our life. On the one hand, we often use APPs or APIs of ASR to generate subtitles and record meetings. On the other hand, smart speaker and self-driving car rely on ASR systems to control AIoT devices. In past few years, there are a lot of works on adversarial examples attacks against ASR systems. By adding a small perturbation to the waveforms, the recognition results make a big difference. In this paper, we describe the development of ASR system, different assumptions of attacks, and how to evaluate these attacks. Next, we introduce the current works on adversarial examples attacks from two attack assumptions: white-box attack and black-box attack. Different from other surveys, we pay more attention to which layer they perturb waveforms in ASR system, the relationship between these attacks, and their implementation methods. We focus on the effect of their works.

Reinforcement Learning (RL) techniques have drawn great attention in many challenging tasks, but their performance deteriorates dramatically when applied to real-world problems. Various methods, such as domain randomization, have been proposed to deal with such situations by training agents under different environmental setups, and therefore they can be generalized to different environments during deployment. However, they usually do not incorporate the underlying environmental factor information that the agents interact with properly and thus can be overly conservative when facing changes in the surroundings. In this paper, we first formalize the task of adapting to changing environmental dynamics in RL as a generalization problem using Contextual Markov Decision Processes (CMDPs). We then propose the Asymmetric Actor-Critic in Contextual RL (AACC) as an end-to-end actor-critic method to deal with such generalization tasks. We demonstrate the essential improvements in the performance of AACC over existing baselines experimentally in a range of simulated environments.

Autonomous navigation in highly populated areas remains a challenging task for robots because of the difficulty in guaranteeing safe interactions with pedestrians in unstructured situations. In this work, we present a crowd navigation control framework that delivers continuous obstacle avoidance and post-contact control evaluated on an autonomous personal mobility vehicle. We propose evaluation metrics for accounting efficiency, controller response and crowd interactions in natural crowds. We report the results of over 110 trials in different crowd types: sparse, flows, and mixed traffic, with low- (< 0.15 ppsm), mid- (< 0.65 ppsm), and high- (< 1 ppsm) pedestrian densities. We present comparative results between two low-level obstacle avoidance methods and a baseline of shared control. Results show a 10% drop in relative time to goal on the highest density tests, and no other efficiency metric decrease. Moreover, autonomous navigation showed to be comparable to shared-control navigation with a lower relative jerk and significantly higher fluency in commands indicating high compatibility with the crowd. We conclude that the reactive controller fulfils a necessary task of fast and continuous adaptation to crowd navigation, and it should be coupled with high-level planners for environmental and situational awareness.

Machine learning based malware detection techniques rely on grayscale images of malware and tends to classify malware based on the distribution of textures in graycale images. Albeit the advancement and promising results shown by machine learning techniques, attackers can exploit the vulnerabilities by generating adversarial samples. Adversarial samples are generated by intelligently crafting and adding perturbations to the input samples. There exists majority of the software based adversarial attacks and defenses. To defend against the adversaries, the existing malware detection based on machine learning and grayscale images needs a preprocessing for the adversarial data. This can cause an additional overhead and can prolong the real-time malware detection. So, as an alternative to this, we explore RRAM (Resistive Random Access Memory) based defense against adversaries. Therefore, the aim of this thesis is to address the above mentioned critical system security issues. The above mentioned challenges are addressed by demonstrating proposed techniques to design a secure and robust cognitive system. First, a novel technique to detect stealthy malware is proposed. The technique uses malware binary images and then extract different features from the same and then employ different ML-classifiers on the dataset thus obtained. Results demonstrate that this technique is successful in differentiating classes of malware based on the features extracted. Secondly, I demonstrate the effects of adversarial attacks on a reconfigurable RRAM-neuromorphic architecture with different learning algorithms and device characteristics. I also propose an integrated solution for mitigating the effects of the adversarial attack using the reconfigurable RRAM architecture.