Machine learning (ML) and deep learning systems are being deployed in sensitive applications, but they can fail in multiple ways, impacting the confidentiality, integrity and availability of user data [33]. To date, evasion attacks or inference-time attacks have been studied extensively in image classification [60, 20, 7], speech recognition [9, 49], and cyber security [58, 68, 30]. Still, among the threats machine learning and deep learning systems are vulnerable to, poisoning attacks at training time has recently surfaced as the threat perceived as most potentially dangerous to companies' ML infrastructures [52]. The threat of poisoning attacks becomes even more severe as modern deep learning systems rely on large, diverse datasets, and their size makes it difficult to guarantee the trustworthiness of the training data. In existing poisoning attacks, adversaries can insert a set of corrupted, poisoned data at training time to induce a specific outcome in classification at inference time. Existing poisoning attacks can be classified into: availability attacks [4, 67, 25] in which the overall accuracy of the model is degraded; targeted attacks [27, 51, 59] in which specific test instances are targeted for misclassification; and backdoor attacks [21] in which a backdoor pattern added to testing points induces misclassification. Poisoning attacks range in the amount of knowledge the attacker has about the ML system, with white-box attacks assuming full knowledge, black-box attacks assuming minimal knowledge about the ML system, and gray-box attacks assuming partial knowledge, such as the feature representation or model architecture [59]. The threat models for poisoning attacks defined in the literature rely on strong assumptions on the adversarial capabilities. In both poisoning availability attacks [4, 67, 25] and backdoor attacks [21] the adversary needs to control a relatively large fraction of the training data (e.g., 10% or 20%) to influence the model at inference time.

Over the past few years, deep neural networks have shown excellent results in multiple tasks, however, there is still an increasing need to address the problem of interpretability to improve model transparency, performance, and safety. Achieving eXplainable Artificial Intelligence (XAI) by combining neural networks with continuous logic and multi-criteria decision-making tools is one of the most promising ways to approach this problem: by this combination, the black-box nature of neural models can be reduced. The continuous logic-based neural model uses so-called Squashing activation functions, a parametric family of functions that satisfy natural invariance requirements and contain rectified linear units as a particular case. This work demonstrates the first benchmark tests that measure the performance of Squashing functions in neural networks. Three experiments were carried out to examine their usability and a comparison with the most popular activation functions was made for five different network types. The performance was determined by measuring the accuracy, loss, and time per epoch. These experiments and the conducted benchmarks have proven that the use of Squashing functions is possible and similar in performance to conventional activation functions. Moreover, a further experiment was conducted by implementing nilpotent logical gates to demonstrate how simple classification tasks can be solved successfully and with high performance. The results indicate that due to the embedded nilpotent logical operators and the differentiability of the Squashing function, it is possible to solve classification problems, where other commonly used activation functions fail.

Imbalanced learning (IL), i.e., learning unbiased models from class-imbalanced data, is a challenging problem. Typical IL methods including resampling and reweighting were designed based on some heuristic assumptions. They often suffer from unstable performance, poor applicability, and high computational cost in complex tasks where their assumptions do not hold. In this paper, we introduce a novel ensemble IL framework named MESA. It adaptively resamples the training set in iterations to get multiple classifiers and forms a cascade ensemble model. MESA directly learns the sampling strategy from data to optimize the final metric beyond following random heuristics. Moreover, unlike prevailing meta-learning-based IL solutions, we decouple the model-training and meta-training in MESA by independently train the meta-sampler over task-agnostic meta-data. This makes MESA generally applicable to most of the existing learning models and the meta-sampler can be efficiently applied to new tasks. Extensive experiments on both synthetic and real-world tasks demonstrate the effectiveness, robustness, and transferability of MESA. Our code is available at https://github.com/ZhiningLiu1998/mesa.

The thesis explores the role machine learning methods play in creating intuitive computational models of neural processing. Combined with interpretability techniques, machine learning could replace human modeler and shift the focus of human effort to extracting the knowledge from the ready-made models and articulating that knowledge into intuitive descroptions of reality. This perspective makes the case in favor of the larger role that exploratory and data-driven approach to computational neuroscience could play while coexisting alongside the traditional hypothesis-driven approach. We exemplify the proposed approach in the context of the knowledge representation taxonomy with three research projects that employ interpretability techniques on top of machine learning methods at three different levels of neural organization. The first study (Chapter 3) explores feature importance analysis of a random forest decoder trained on intracerebral recordings from 100 human subjects to identify spectrotemporal signatures that characterize local neural activity during the task of visual categorization. The second study (Chapter 4) employs representation similarity analysis to compare the neural responses of the areas along the ventral stream with the activations of the layers of a deep convolutional neural network. The third study (Chapter 5) proposes a method that allows test subjects to visually explore the state representation of their neural signal in real time. This is achieved by using a topology-preserving dimensionality reduction technique that allows to transform the neural data from the multidimensional representation used by the computer into a two-dimensional representation a human can grasp. The approach, the taxonomy, and the examples, present a strong case for the applicability of machine learning methods to automatic knowledge discovery in neuroscience.

Data Parallelism (DP) and Model Parallelism (MP) are two common paradigms to enable large-scale distributed training of neural networks. Recent trends, such as the improved model performance of deeper and wider neural networks when trained with billions of data points, have prompted the use of hybrid parallelism---a paradigm that employs both DP and MP to scale further parallelization for machine learning. Hybrid training allows compute power to increase, but it runs up against the key bottleneck of communication overhead that hinders scalability. In this paper, we propose a compression framework called Dynamic Communication Thresholding (DCT) for communication-efficient hybrid training. DCT filters the entities to be communicated across the network through a simple hard-thresholding function, allowing only the most relevant information to pass through. For communication efficient DP, DCT compresses the parameter gradients sent to the parameter server during model synchronization, while compensating for the introduced errors with known techniques. For communication efficient MP, DCT incorporates a novel technique to compress the activations and gradients sent across the network during the forward and backward propagation, respectively. This is done by identifying and updating only the most relevant neurons of the neural network for each training sample in the data. Under modest assumptions, we show that the convergence of training is maintained with DCT. We evaluate DCT on natural language processing and recommender system models. DCT reduces overall communication by 20x, improving end-to-end training time on industry scale models by 37%. Moreover, we observe an improvement in the trained model performance, as the induced sparsity is possibly acting as an implicit sparsity based regularization.

Contrastive representation learning has shown to be an effective way of learning representations from unlabeled data. However, much progress has been made in vision domains relying on data augmentations carefully designed using domain knowledge. In this work, we propose i-Mix, a simple yet effective regularization strategy for improving contrastive representation learning in both vision and non-vision domains. We cast contrastive learning as training a non-parametric classifier by assigning a unique virtual class to each data in a batch. Then, data instances are mixed in both the input and virtual label spaces, providing more augmented data during training. In experiments, we demonstrate that i-Mix consistently improves the quality of self-supervised representations across domains, resulting in significant performance gains on downstream tasks. Furthermore, we confirm its regularization effect via extensive ablation studies across model and dataset sizes.

We propose a novel graph cross network (GXN) to achieve comprehensive feature learning from multiple scales of a graph. Based on trainable hierarchical representations of a graph, GXN enables the interchange of intermediate features across scales to promote information flow. Two key ingredients of GXN include a novel vertex infomax pooling (VIPool), which creates multiscale graphs in a trainable manner, and a novel feature-crossing layer, enabling feature interchange across scales. The proposed VIPool selects the most informative subset of vertices based on the neural estimation of mutual information between vertex features and neighborhood features. The intuition behind is that a vertex is informative when it can maximally reflect its neighboring information. The proposed feature-crossing layer fuses intermediate features between two scales for mutual enhancement by improving information flow and enriching multiscale features at hidden layers. The cross shape of the feature-crossing layer distinguishes GXN from many other multiscale architectures. Experimental results show that the proposed GXN improves the classification accuracy by 2.12% and 1.15% on average for graph classification and vertex classification, respectively. Based on the same network, the proposed VIPool consistently outperforms other graph-pooling methods.

We study the problem of learning data representations that are private yet informative, i.e., providing information about intended "ally" targets while obfuscating sensitive "adversary" attributes. We propose a novel framework, Exclusion-Inclusion Generative Adversarial Network (EIGAN), that generalizes existing adversarial privacy-preserving representation learning (PPRL) approaches to generate data encodings that account for multiple possibly overlapping ally and adversary targets. Preserving privacy is even more difficult when the data is collected across multiple distributed nodes, which for privacy reasons may not wish to share their data even for PPRL training. Thus, learning such data representations at each node in a distributed manner (i.e., without transmitting source data) is of particular importance. This motivates us to develop D-EIGAN, the first distributed PPRL method, based on federated learning with fractional parameter sharing to account for communication resource limitations. We theoretically analyze the behavior of adversaries under the optimal EIGAN and D-EIGAN encoders and consider the impact of dependencies among ally and adversary tasks on the encoder performance. Our experiments on real-world and synthetic datasets demonstrate the advantages of EIGAN encodings in terms of accuracy, robustness, and scalability; in particular, we show that EIGAN outperforms the previous state-of-the-art by a significant accuracy margin (47% improvement). The experiments further reveal that D-EIGAN's performance is consistent with EIGAN under different node data distributions and is resilient to communication constraints.

Recently, flood damage has become a social problem owing to unexperienced weather conditions arising from climate change. An immediate response to heavy rain and high water levels is important for the mitigation of casualties and economic losses and also for rapid recovery. Spatiotemporal precipitation forecasts may enhance the accuracy of dam inflow prediction, more than 6 hours forward for flood damage mitigation. This paper proposes a rain-code approach for spatiotemporal precipitation forecasting. We propose a novel rainy feature that represents a temporal rainy process using multi-frame fusion for timestep reduction. We perform rain-code studies with various term ranges based on spatiotemporal precipitation forecasting using the standard ConvLSTM. We applied to a dam region within the Japanese rainy term hourly precipitation data, under 2006 to 2019 approximately 127 thousands hours, every year from May to October. We apply the radar analysis hourly data on the central broader region with an area of 136 x 148 km2 , based on new data fusion rain code with multi-frame sequences. Finally we got some evidences and capabilities for strengthen forecasting range.

Counterfactual learning for dealing with missing-not-at-random data (MNAR) is an intriguing topic in the recommendation literature since MNAR data are ubiquitous in modern recommender systems. Missing-at-random (MAR) data, namely randomized controlled trials (RCTs), are usually required by most previous counterfactual learning methods for debiasing learning. However, the execution of RCTs is extraordinarily expensive in practice. To circumvent the use of RCTs, we build an information-theoretic counterfactual variational information bottleneck (CVIB), as an alternative for debiasing learning without RCTs. By separating the task-aware mutual information term in the original information bottleneck Lagrangian into factual and counterfactual parts, we derive a contrastive information loss and an additional output confidence penalty, which facilitates balanced learning between the factual and counterfactual domains. Empirical evaluation on real-world datasets shows that our CVIB significantly enhances both shallow and deep models, which sheds light on counterfactual learning in recommendation that goes beyond RCTs.