Systems used here must provide comprehensible and transparent information about their decisions. Especially for patients, who are mostly no healthcare experts, comprehensible information is extremely important to understand diagnoses and treatment options (e.g., [2, 3]). To support more transparent Artificial Intelligence (AI) applications, approaches for Explainable Artificial Intelligence (XAI) are an ongoing topic of high interest [4]. Especially in the field of computer vision, a common strategy to achieve this kind of transparency is the creation of saliency maps that highlight areas in the input that were important for the decision of the AI system. The problem with those explanation strategies is, that they require the user of the XAI system to perform an additional thought process: Having the information what areas were important for a certain decision inevitably leads to the question why these areas were of importance. Especially in scenarios where relevant differences in the input data are originating from textural information rather than spatial information of certain objects, it becomes clear that the raw information about where important areas are is not always sufficient. One XAI approach that goes another way to avoid the aforementioned problems, are Counterfactual Explanations. Counterfactual explanations try to help to understand why the actual decision was made instead of another one, by creating a slightly modified version of the input which results in another decision of the AI [5, 6]. Creating such a slightly modified input that changes the model's prediction is by no means a trivial task.

Recent studies have indicated that Graph Convolutional Networks (GCNs) act as a low pass filter in spectral domain and encode smoothed node representations. In this paper, we consider their opposite, namely Graph Deconvolutional Networks (GDNs) that reconstruct graph signals from smoothed node representations. We motivate the design of Graph Deconvolutional Networks via a combination of inverse filters in spectral domain and de-noising layers in wavelet domain, as the inverse operation results in a high pass filter and may amplify the noise. Based on the proposed GDN, we further propose a graph autoencoder framework that first encodes smoothed graph representations with GCN and then decodes accurate graph signals with GDN. We demonstrate the effectiveness of the proposed method on several tasks including unsupervised graph-level representation, social recommendation and graph generation. Autoencoders have demonstrated excellent performance on tasks such as unsupervised representation learning (Bengio, 2009) and de-noising (Vincent et al., 2010). Recently, several studies (Zeiler & Fergus, 2014; Long et al., 2015) have demonstrated that the performance of autoencoders can be further improved by encoding with Convolutional Networks and decoding with Deconvolutional Networks (Zeiler et al., 2010). Notably, Noh et al. (2015) present a novel symmetric architecture that provides a bottom-up mapping from input signals to latent hierarchical feature space with {convolution, pooling} operations and then maps the latent representation back to the input space with {deconvolution, unpooling} operations. While this architecture has been successful when processing features with structures existed in the Euclidean space (e.g., images), recently there has been a surging interest in applying such a framework on non-Euclidean data like graphs.

The presence of label noise often misleads the training of deep neural networks. Departing from the recent literature which largely assumes the label noise rate is only determined by the true class, the errors in human-annotated labels are more likely to be dependent on the difficulty levels of tasks, resulting in settings with instance-dependent label noise. We show theoretically that the heterogeneous instance-dependent label noise is effectively down-weighting the examples with higher noise rates in a non-uniform way and thus causes imbalances, rendering the strategy of directly applying methods for class-dependent label noise questionable. In this paper, we propose and study the potentials of a second-order approach that leverages the estimation of several covariance terms defined between the instance-dependent noise rates and the Bayes optimal label. We show that this set of second-order information successfully captures the induced imbalances. We further proceed to show that with the help of the estimated second-order information, we identify a new loss function whose expected risk of a classifier under instance-dependent label noise can be shown to be equivalent to a new problem with only class-dependent label noise. This fact allows us to develop effective loss functions to correctly evaluate models. We provide an efficient procedure to perform the estimations without accessing either ground truth labels or prior knowledge of the noise rates. Experiments on CIFAR10 and CIFAR100 with synthetic instance-dependent label noise and Clothing1M with real-world human label noise verify our approach.

In the current context of Big Data, the nature of many forecasting problems has changed from predicting isolated time series to predicting many time series from similar sources. This has opened up the opportunity to develop competitive global forecasting models that simultaneously learn from many time series. But, it still remains unclear when global forecasting models can outperform the univariate benchmarks, especially along the dimensions of the homogeneity/heterogeneity of series, the complexity of patterns in the series, the complexity of forecasting models, and the lengths/number of series. Our study attempts to address this problem through investigating the effect from these factors, by simulating a number of datasets that have controllable time series characteristics. Specifically, we simulate time series from simple data generating processes (DGP), such as Auto Regressive (AR) and Seasonal AR, to complex DGPs, such as Chaotic Logistic Map, Self-Exciting Threshold Auto-Regressive, and Mackey-Glass Equations. The data heterogeneity is introduced by mixing time series generated from several DGPs into a single dataset. The lengths and the number of series in the dataset are varied in different scenarios. We perform experiments on these datasets using global forecasting models including Recurrent Neural Networks (RNN), Feed-Forward Neural Networks, Pooled Regression (PR) models and Light Gradient Boosting Models (LGBM), and compare their performance against standard statistical univariate forecasting techniques. Our experiments demonstrate that when trained as global forecasting models, techniques such as RNNs and LGBMs, which have complex non-linear modelling capabilities, are competitive methods in general under challenging forecasting scenarios such as series having short lengths, datasets with heterogeneous series and having minimal prior knowledge of the patterns of the series.

Labeling training examples at scale is a perennial challenge in machine learning. Self-supervision methods compensate for the lack of direct supervision by leveraging prior knowledge to automatically generate noisy labeled examples. Deep probabilistic logic (DPL) is a unifying framework for self-supervised learning that represents unknown labels as latent variables and incorporates diverse self-supervision using probabilistic logic to train a deep neural network end-to-end using variational EM. While DPL is successful at combining pre-specified self-supervision, manually crafting self-supervision to attain high accuracy may still be tedious and challenging. In this paper, we propose Self-Supervised Self-Supervision (S4), which adds to DPL the capability to learn new self-supervision automatically. Starting from an initial "seed," S4 iteratively uses the deep neural network to propose new self supervision. These are either added directly (a form of structured self-training) or verified by a human expert (as in feature-based active learning). Experiments show that S4 is able to automatically propose accurate self-supervision and can often nearly match the accuracy of supervised methods with a tiny fraction of the human effort.

After decades of space travel, low Earth orbit is a junkyard of discarded rocket bodies, dead satellites, and millions of pieces of debris from collisions and explosions. Objects in high enough altitudes do not re-enter and burn up in the atmosphere, but stay in orbit around Earth for a long time. With a speed of 28,000 km/h, collisions in these orbits can generate fragments and potentially trigger a cascade of more collisions known as the Kessler syndrome. This could pose a planetary challenge, because the phenomenon could escalate to the point of hindering future space operations and damaging satellite infrastructure critical for space and Earth science applications. As commercial entities place mega-constellations of satellites in orbit, the burden on operators conducting collision avoidance manoeuvres will increase. For this reason, development of automated tools that predict potential collision events (conjunctions) is critical. We introduce a Bayesian deep learning approach to this problem, and develop recurrent neural network architectures (LSTMs) that work with time series of conjunction data messages (CDMs), a standard data format used by the space community. We show that our method can be used to model all CDM features simultaneously, including the time of arrival of future CDMs, providing predictions of conjunction event evolution with associated uncertainties.

Non-convex optimization problems are challenging to solve; the success and computational expense of a gradient descent algorithm or variant depend heavily on the initialization strategy. Often, either random initialization is used or initialization rules are carefully designed by exploiting the nature of the problem class. As a simple alternative to hand-crafted initialization rules, we propose an approach for learning "good" initialization rules from previous solutions. We provide theoretical guarantees that establish conditions that are sufficient in all cases and also necessary in some under which our approach performs better than random initialization. We apply our methodology to various non-convex problems such as generating adversarial examples, generating post hoc explanations for black-box machine learning models, and allocating communication spectrum, and show consistent gains over other initialization techniques.

There is an urgent need to build models to tackle Indoor Air Quality issue. Since the model should be accurate and fast, Reduced Order Modelling technique is used to reduce the dimensionality of the problem. The accuracy of the model, that represent a dynamic system, is improved integrating real data coming from sensors using Data Assimilation techniques. In this paper, we formulate a new methodology called Latent Assimilation that combines Data Assimilation and Machine Learning. We use a Convolutional neural network to reduce the dimensionality of the problem, a Long-Short-Term-Memory to build a surrogate model of the dynamic system and an Optimal Interpolated Kalman Filter to incorporate real data. Experimental results are provided for CO2 concentration within an indoor space. This methodology can be used for example to predict in real-time the load of virus, such as the SARS-COV-2, in the air by linking it to the concentration of CO2.

Good mental health is important, and early research suggests AI might help diagnose that health for people who are under a particularly heavy strain. Scientists have developed a combination of deep learning algorithms that use eye tracking to gauge the mental health of cancer patients after surgery. Ideally, this would help spot patients likely to be suffering from anxiety or depression when a human can't perform an initial psychological assessment. The system uses a mix of a convolutional neural network and long short-term memory algorithms to study the eye movements of patients wearing tracking glasses (in this case, Tobii Pro 2 glasses) while they contemplate artwork. The AI used the gaze and pupil position data from those glasses to determine how likely someone was to raise concerns on established hope, anxiety and mental wellbeing questionnaires they would fill out later.

Neural networks with deep architectures have demonstrated significant performance improvements in computer vision, speech recognition, and natural language processing. The challenges in information retrieval (IR), however, are different from these other application areas. A common form of IR involves ranking of documents -- or short passages -- in response to keyword-based queries. Effective IR systems must deal with query-document vocabulary mismatch problem, by modeling relationships between different query and document terms and how they indicate relevance. Models should also consider lexical matches when the query contains rare terms -- such as a person's name or a product model number -- not seen during training, and to avoid retrieving semantically related but irrelevant results. In many real-life IR tasks, the retrieval involves extremely large collections -- such as the document index of a commercial Web search engine -- containing billions of documents. Efficient IR methods should take advantage of specialized IR data structures, such as inverted index, to efficiently retrieve from large collections. Given an information need, the IR system also mediates how much exposure an information artifact receives by deciding whether it should be displayed, and where it should be positioned, among other results. Exposure-aware IR systems may optimize for additional objectives, besides relevance, such as parity of exposure for retrieved items and content publishers. In this thesis, we present novel neural architectures and methods motivated by the specific needs and challenges of IR tasks.