Bayesian deep learning offers a principled way to address many issues concerning safety of artificial intelligence (AI), such as model uncertainty,model interpretability, and prediction bias. However, due to the lack of efficient Monte Carlo algorithms for sampling from the posterior of deep neural networks (DNNs), Bayesian deep learning has not yet powered our AI system. We propose a class of adaptive stochastic gradient Markov chain Monte Carlo (SGMCMC) algorithms, where the drift function is biased to enhance escape from saddle points and the bias is adaptively adjusted according to the gradient of past samples. We establish the convergence of the proposed algorithms under mild conditions, and demonstrate via numerical examples that the proposed algorithms can significantly outperform the existing SGMCMC algorithms, such as stochastic gradient Langevin dynamics (SGLD), stochastic gradient Hamiltonian Monte Carlo (SGHMC) and preconditioned SGLD, in both simulation and optimization tasks.

Risk specialists are trying to understand risk better and use complex models for risk assessment, while many risks are not yet well understood. The lack of empirical data and complex causal and outcome relationships make it difficult to estimate the degree to which certain risk types are exposed. Traditional risk models are based on classical set theory. In comparison, fuzzy logic models are built on fuzzy set theory and are useful for analyzing risks with insufficient knowledge or inaccurate data. Fuzzy logic systems help to make large-scale risk management frameworks more simple. For risks that do not have an appropriate probability model, a fuzzy logic system can help model the cause and effect relationships, assess the level of risk exposure, rank key risks in a consistent way, and consider available data and experts'opinions. Besides, in fuzzy logic systems, some rules explicitly explain the connection, dependence, and relationships between model factors. This can help identify risk mitigation solutions. Resources can be used to mitigate risks with very high levels of exposure and relatively low hedging costs. Fuzzy set and fuzzy logic models can be used with Bayesian and other types of method recognition and decision models, including artificial neural networks and decision tree models. These developed models have the potential to solve difficult risk assessment problems. This research paper explores areas in which fuzzy logic models can be used to improve risk assessment and risk decision making. We will discuss the methodology, framework, and process of using fuzzy logic systems in risk assessment.

One of the most common and important destructive attacks on the victim system is Advanced Persistent Threat (APT)-attack. The APT attacker can achieve his hostile goals by obtaining information and gaining financial benefits regarding the infrastructure of a network. One of the solutions to detect a secret APT attack is using network traffic. Due to the nature of the APT attack in terms of being on the network for a long time and the fact that the network may crash because of high traffic, it is difficult to detect this type of attack. Hence, in this study, machine learning methods such as C5.0 decision tree, Bayesian network and deep neural network are used for timely detection and classification of APT-attacks on the NSL-KDD dataset. Moreover, 10-fold cross validation method is used to experiment these models. As a result, the accuracy (ACC) of the C5.0 decision tree, Bayesian network and 6-layer deep learning models is obtained as 95.64%, 88.37% and 98.85%, respectively, and also, in terms of the important criterion of the false positive rate (FPR), the FPR value for the C5.0 decision tree, Bayesian network and 6-layer deep learning models is obtained as 2.56, 10.47 and 1.13, respectively. Other criterions such as sensitivity, specificity, accuracy, false negative rate and F-measure are also investigated for the models, and the experimental results show that the deep learning model with automatic multi-layered extraction of features has the best performance for timely detection of an APT-attack comparing to other classification models.

Feature selection aims to select a subset of features to optimize the performances of downstream predictive tasks. Recently, multi-agent reinforced feature selection (MARFS) has been introduced to automate feature selection, by creating agents for each feature to select or deselect corresponding features. Although MARFS enjoys the automation of the selection process, MARFS suffers from not just the data complexity in terms of contents and dimensionality, but also the exponentially-increasing computational costs with regard to the number of agents. The raised concern leads to a new research question: Can we simplify the selection process of agents under reinforcement learning context so as to improve the efficiency and costs of feature selection? To address the question, we develop a single-agent reinforced feature selection approach integrated with restructured choice strategy. Specifically, the restructured choice strategy includes: 1) we exploit only one single agent to handle the selection task of multiple features, instead of using multiple agents. 2) we develop a scanning method to empower the single agent to make multiple selection/deselection decisions in each round of scanning. 3) we exploit the relevance to predictive labels of features to prioritize the scanning orders of the agent for multiple features. 4) we propose a convolutional auto-encoder algorithm, integrated with the encoded index information of features, to improve state representation. 5) we design a reward scheme that take into account both prediction accuracy and feature redundancy to facilitate the exploration process. Finally, we present extensive experimental results to demonstrate the efficiency and effectiveness of the proposed method.

This paper considers the problem of estimating the information leakage of a system in the black-box scenario. It is assumed that the system's internals are unknown to the learner, or anyway too complicated to analyze, and the only available information are pairs of input-output data samples, possibly obtained by submitting queries to the system or provided by a third party. Previous research has mainly focused on counting the frequencies to estimate the input-output conditional probabilities (referred to as frequentist approach), however this method is not accurate when the domain of possible outputs is large. To overcome this difficulty, the estimation of the Bayes error of the ideal classifier was recently investigated using Machine Learning (ML) models and it has been shown to be more accurate thanks to the ability of those models to learn the input-output correspondence. However, the Bayes vulnerability is only suitable to describe one-try attacks. A more general and flexible measure of leakage is the g-vulnerability, which encompasses several different types of adversaries, with different goals and capabilities. In this paper, we propose a novel approach to perform black-box estimation of the g-vulnerability using ML. A feature of our approach is that it does not require to estimate the conditional probabilities, and that it is suitable for a large class of ML algorithms. First, we formally show the learnability for all data distributions. Then, we evaluate the performance via various experiments using k-Nearest Neighbors and Neural Networks. Our results outperform the frequentist approach when the observables domain is large.

Using attention weights to identify information that is important for models' decision making is a popular approach to interpret attention-based neural networks, which is commonly realized via creating a heat-map for every single document based on attention weights. However, this interpretation method is fragile. In this paper, we propose a corpus-level explanation approach, which aims to capture causal relationships between keywords and model predictions via learning importance of keywords for predicted labels across a training corpus based on attention weights. Using this idea as the fundamental building block, we further propose a concept-based explanation method that can automatically learn higher-level concepts and their importance to model prediction task. Our concept-based explanation method is built upon a novel Abstraction-Aggregation Network, which can automatically cluster important keywords during an end-to-end training process. We apply these methods to the document classification task and show that they are powerful in extracting semantically meaningful keywords and concepts. Our consistency analysis results based on an attention-based Na\"ive Bayes Classifier also demonstrate these keywords and concepts are important for model predictions.

The project management field has the imperative to increase the project probability of success. Experts have developed several project management maturity models to assets and improve the project outcome. However, the current literature lacks of models allowing correlating the measured maturity and the expected probability of success. This paper uses the characteristics of Bayesian networks to formalize experts' knowledge and to extract knowledge from a project overcost database. It develops a method to estimate the impact of project management maturity on the risk of project overcost. A general framework is presented. An industrial case is used to illustrate the application of the method.

SHAP explanations are a popular feature-attribution mechanism for explainable AI. They use game-theoretic notions to measure the influence of individual features on the prediction of a machine learning model. Despite a lot of recent interest from both academia and industry, it is not known whether SHAP explanations of common machine learning models can be computed efficiently. In this paper, we establish the complexity of computing the SHAP explanation in three important settings. First, we consider fully-factorized data distributions, and show that the complexity of computing the SHAP explanation is the same as the complexity of computing the expected value of the model. This fully-factorized setting is often used to simplify the SHAP computation, yet our results show that the computation can be intractable for commonly used models such as logistic regression. Going beyond fully-factorized distributions, we show that computing SHAP explanations is already intractable for a very simple setting: computing SHAP explanations of trivial classifiers over naive Bayes distributions. Finally, we show that even computing SHAP over the empirical distribution is #P-hard.

Machine learning systems are increasingly being used to make impactful decisions such as loan applications and criminal justice risk assessments, and as such, ensuring fairness of these systems is critical. This is often challenging as the labels in the data are biased. This paper studies learning fair probability distributions from biased data by explicitly modeling a latent variable that represents a hidden, unbiased label. In particular, we aim to achieve demographic parity by enforcing certain independencies in the learned model. We also show that group fairness guarantees are meaningful only if the distribution used to provide those guarantees indeed captures the real-world data. In order to closely model the data distribution, we employ probabilistic circuits, an expressive and tractable probabilistic model, and propose an algorithm to learn them from incomplete data. We evaluate our approach on a synthetic dataset in which observed labels indeed come from fair labels but with added bias, and demonstrate that the fair labels are successfully retrieved. Moreover, we show on real-world datasets that our approach not only is a better model than existing methods of how the data was generated but also achieves competitive accuracy.

Inducing a meaningful structural representation from one or a set of dialogues is a crucial but challenging task in computational linguistics. Advancement made in this area is critical for dialogue system design and discourse analysis. It can also be extended to solve grammatical inference. In this work, we propose to incorporate structured attention layers into a Variational Recurrent Neural Network (VRNN) model with discrete latent states to learn dialogue structure in an unsupervised fashion. Compared to a vanilla VRNN, structured attention enables a model to focus on different parts of the source sentence embeddings while enforcing a structural inductive bias. Experiments show that on two-party dialogue datasets, VRNN with structured attention learns semantic structures that are similar to templates used to generate this dialogue corpus. While on multi-party dialogue datasets, our model learns an interactive structure demonstrating its capability of distinguishing speakers or addresses, automatically disentangling dialogues without explicit human annotation.