We provide a novel approach to synthesize controllers for nonlinear continuous dynamical systems with control against safety properties. The controllers are based on neural networks (NNs). To certify the safety property we utilize barrier functions, which are represented by NNs as well. We train the controller-NN and barrier-NN simultaneously, achieving a verification-in-the-loop synthesis. We provide a prototype tool nncontroller with a number of case studies. The experiment results confirm the feasibility and efficacy of our approach.

Convolutional neural networks (CNN) are known to learn an image representation that captures concepts relevant to the task, but do so in an implicit way that hampers model interpretability. However, one could argue that such a representation is hidden in the neurons and can be made explicit by teaching the model to recognize semantically interpretable attributes that are present in the scene. We call such an intermediate layer a \emph{semantic bottleneck}. Once the attributes are learned, they can be re-combined to reach the final decision and provide both an accurate prediction and an explicit reasoning behind the CNN decision. In this paper, we look into semantic bottlenecks that capture context: we want attributes to be in groups of a few meaningful elements and participate jointly to the final decision. We use a two-layer semantic bottleneck that gathers attributes into interpretable, sparse groups, allowing them contribute differently to the final output depending on the context. We test our contextual semantic interpretable bottleneck (CSIB) on the task of landscape scenicness estimation and train the semantic interpretable bottleneck using an auxiliary database (SUN Attributes). Our model yields in predictions as accurate as a non-interpretable baseline when applied to a real-world test set of Flickr images, all while providing clear and interpretable explanations for each prediction.

This paper reviews the current state of the art in Artificial Intelligence (AI) technologies and applications in the context of the creative industries. A brief background of AI, and specifically Machine Learning (ML) algorithms, is provided including Convolutional Neural Network (CNNs), Generative Adversarial Networks (GANs), Recurrent Neural Networks (RNNs) and Deep Reinforcement Learning (DRL). We categorise creative applications into five groups related to how AI technologies are used: i) content creation, ii) information analysis, iii) content enhancement and post production workflows, iv) information extraction and enhancement, and v) data compression. We critically examine the successes and limitations of this rapidly advancing technology in each of these areas. We further differentiate between the use of AI as a creative tool and its potential as a creator in its own right. We foresee that, in the near future, machine learning-based AI will be adopted widely as a tool or collaborative assistant for creativity. In contrast, we observe that the successes of machine learning in domains with fewer constraints, where AI is the `creator', remain modest. The potential of AI (or its developers) to win awards for its original creations in competition with human creatives is also limited, based on contemporary technologies. We therefore conclude that, in the context of creative industries, maximum benefit from AI will be derived where its focus is human centric -- where it is designed to augment, rather than replace, human creativity.

Imagine a scenario where we are trying to detect a shooting using microphones deployed in a city: shooting is a situation of interest that we want to identify from a high-throughput (audio) data stream. Complex Event Processing (CEP) is a type of approach aimed at detecting such situations of interest, called complex events, from a data stream using a set of rules. These rules are defined on atomic pieces of information from the data stream, which we call events--or simple events, for clarity. Complex events can be formed from multiple simple events. For instance, shooting might start when multiple instances of the simple event gunshot occur. For simplicity, we can assume that when we start to detect siren events, authorities have arrived and the situation is being dealt with, which would conclude the complex event. Using the raw data stream implies that usually we cannot directly write declarative rules on that data, as it would imply that we need to process that raw data using symbolic rules; though theoretically possible, this is hardly recommended. Using a machine learning algorithm such a neural network trained with back-propagation is also infeasible, as it will need to simultaneously learn to understand the simple events within the data stream, and the interrelationship between such events to compose a complex event. While possible, the sparsity of data makes this a hard problem to solve.

We leverage what are typically considered the worst qualities of deep learning algorithms - high computational cost, requirement for large data, no explainability, high dependence on hyper-parameter choice, overfitting, and vulnerability to adversarial perturbations - in order to create a method for the secure and efficient training of remotely deployed neural networks over unsecured channels.

Multi-Task Learning (MTL) has emerged as a promising approach for transferring learned knowledge across different tasks. However, multi-task learning must deal with challenges such as: overfitting to low resource tasks, catastrophic forgetting, and negative task transfer, or learning interference. Additionally, in Natural Language Processing (NLP), MTL alone has typically not reached the performance level possible through per-task fine-tuning of pretrained models. However, many fine-tuning approaches are both parameter inefficient, e.g. potentially involving one new model per task, and highly susceptible to losing knowledge acquired during pretraining. We propose a novel transformer based architecture consisting of a new conditional attention mechanism as well as a set of task conditioned modules that facilitate weight sharing. Through this construction we achieve more efficient parameter sharing and mitigate forgetting by keeping half of the weights of a pretrained model fixed. We also use a new multi-task data sampling strategy to mitigate the negative effects of data imbalance across tasks. Using this approach we are able to surpass single-task fine-tuning methods while being parameter and data efficient. With our base model, we attain 2.2% higher performance compared to a full fine-tuned BERT large model on the GLUE benchmark, adding only 5.6% more trained parameters per task (whereas naive fine-tuning potentially adds 100% of the trained parameters per task) and needing only 64.6% of the data. We show that a larger variant of our single multi-task model approach performs competitively across 26 NLP tasks and yields state-of-the-art results on a number of test and development sets.

Time series forecasting involves collecting and analyzing past observations to develop a model to extrapolate such observations into the future. Forecasting of future events is important in many fields to support decision making as it contributes to reducing the future uncertainty. We propose explainable boosted linear regression (EBLR) algorithm for time series forecasting, which is an iterative method that starts with a base model, and explains the model's errors through regression trees. At each iteration, the path leading to highest error is added as a new variable to the base model. In this regard, our approach can be considered as an improvement over general time series models since it enables incorporating nonlinear features by residuals explanation. More importantly, use of the single rule that contributes to the error most allows for interpretable results. The proposed approach extends to probabilistic forecasting through generating prediction intervals based on the empirical error distribution. We conduct a detailed numerical study with EBLR and compare against various other approaches. We observe that EBLR substantially improves the base model performance through extracted features, and provide a comparable performance to other well established approaches. The interpretability of the model predictions and high predictive accuracy of EBLR makes it a promising method for time series forecasting.

Being able to interpret a machine learning model is a crucial task in many applications of machine learning. Specifically, local interpretability is important in determining why a model makes particular predictions. Despite the recent focus on AI interpretability, there has been a lack of research in local interpretability methods for time series forecasting while the few interpretable methods that exist mainly focus on time series classification tasks. In this study, we propose two novel evaluation metrics for time series forecasting: Area Over the Perturbation Curve for Regression and Ablation Percentage Threshold. These two metrics can measure the local fidelity of local explanation models. We extend the theoretical foundation to collect experimental results on two popular datasets, \textit{Rossmann sales} and \textit{electricity}. Both metrics enable a comprehensive comparison of numerous local explanation models and find which metrics are more sensitive. Lastly, we provide heuristical reasoning for this analysis.

In decentralized learning, data is distributed among local clients which collaboratively train a shared prediction model using secure aggregation. To preserve the privacy of the clients, modern decentralized learning paradigms require each client to maintain a private local training data set and only upload their summarized model updates to the server. However, this can quickly lead to a degenerate model and collapse in performance when corrupted updates (e.g., adversarial manipulations) are aggregated at the server. In this work, we present a robust decentralized learning framework, Decent_BVA, using bias-variance based adversarial training via asymmetrical communications between each client and the server. The experiments are conducted on neural networks with cross-entropy loss. Nevertheless, the proposed framework allows the use of various classification loss functions (e.g., cross-entropy loss, mean squared error loss) where the gradients of the bias and variance are tractable to be estimated from local clients' models. In this case, any gradient-based adversarial training strategies could be used by taking the bias-variance oriented adversarial examples into consideration, e.g., bias-variance based FGSM and PGD proposed in this paper. Experiments show that Decent_BVA is robust to the classical adversarial attacks when the level of corruption is high while being competitive compared with conventional decentralized learning in terms of the model's accuracy and efficiency.

Concentration inequalities have been widely used in machine learning theory. Model selection techniques, for instance, relies heavily on concentration inequality [Massart, 2007]. They have also been used for high dimensional procedures [Bickel et al., 2009, Alquier et al., 2020] or for studying different machine learning framework, such as time series prediction[Kuznetsov and Mohri, 2015], online machine learning[Sanchez-Perez, 2015] or classification problems [Freund et al., 2004]. Many concentration inequalities has been proposed for different framework and different hypothesis. An interested reader may read [Boucheron et al., 2013] for an overview on stationary concentration inequalities.