Improving Generalization by Controlling Label-Noise Information in Neural Network Weights

In the presence of noisy or incorrect labels, neural networks have the undesirable tendency to memorize information about the noise. Standard regularization techniques such as dropout, weight decay or data augmentation sometimes help, but do not prevent this behavior. If one considers neural network weights as random variables that depend on the data and stochasticity of training, the amount of memorized information can be quantified with the Shannon mutual information between weights and the vector of all training labels given inputs, $I(w : \mathbf{y} \mid \mathbf{x})$. We show that for any training algorithm, low values of this term correspond to reduction in memorization of label-noise and better generalization bounds. To obtain these low values, we propose training algorithms that employ an auxiliary network that predicts gradients in the final layers of a classifier without accessing labels. We illustrate the effectiveness of our approach on versions of MNIST, CIFAR-10, and CIFAR-100 corrupted with various noise models, and on a large-scale dataset Clothing1M that has noisy labels.

Variational Inference for Quantifying Inter-observer Variability in Segmentation of Anatomical Structures

Lesions or organ boundaries visible through medical imaging data are often ambiguous, thus resulting in significant variations in multi-reader delineations, i.e., the source of aleatoric uncertainty. In particular, quantifying the inter-observer variability of manual annotations with Magnetic Resonance (MR) Imaging data plays a crucial role in establishing a reference standard for various diagnosis and treatment tasks. Most segmentation methods, however, simply model a mapping from an image to its single segmentation map and do not take the disagreement of annotators into consideration. In order to account for inter-observer variability, without sacrificing accuracy, we propose a novel variational inference framework to model the distribution of plausible segmentation maps, given a specific MR image, which explicitly represents the multi-reader variability. Specifically, we resort to a latent vector to encode the multi-reader variability and counteract the inherent information loss in the imaging data. Then, we apply a variational autoencoder network and optimize its evidence lower bound (ELBO) to efficiently approximate the distribution of the segmentation map, given an MR image. Experimental results, carried out with the QUBIQ brain growth MRI segmentation datasets with seven annotators, demonstrate the effectiveness of our approach.

Learning with Privileged Information for Efficient Image Super-Resolution

Convolutional neural networks (CNNs) have allowed remarkable advances in single image super-resolution (SISR) over the last decade. Most SR methods based on CNNs have focused on achieving performance gains in terms of quality metrics, such as PSNR and SSIM, over classical approaches. They typically require a large amount of memory and computational units. FSRCNN, consisting of few numbers of convolutional layers, has shown promising results, while using an extremely small number of network parameters. We introduce in this paper a novel distillation framework, consisting of teacher and student networks, that allows to boost the performance of FSRCNN drastically. To this end, we propose to use ground-truth high-resolution (HR) images as privileged information. The encoder in the teacher learns the degradation process, subsampling of HR images, using an imitation loss. The student and the decoder in the teacher, having the same network architecture as FSRCNN, try to reconstruct HR images. Intermediate features in the decoder, affordable for the student to learn, are transferred to the student through feature distillation. Experimental results on standard benchmarks demonstrate the effectiveness and the generalization ability of our framework, which significantly boosts the performance of FSRCNN as well as other SR methods. Our code and model are available online: https://cvlab.yonsei.ac.kr/projects/PISR.

Data-Driven Deep Learning Based Hybrid Beamforming for Aerial Massive MIMO-OFDM Systems with Implicit CSI

In an aerial hybrid massive multiple-input multiple-output (MIMO) and orthogonal frequency division multiplexing (OFDM) system, how to design a spectral-efficient broadband multi-user hybrid beamforming with a limited pilot and feedback overhead is challenging. To this end, by modeling the key transmission modules as an end-to-end (E2E) neural network, this paper proposes a data-driven deep learning (DL)-based unified hybrid beamforming framework for both the time division duplex (TDD) and frequency division duplex (FDD) systems with implicit channel state information (CSI). For TDD systems, the proposed DL-based approach jointly models the uplink pilot combining and downlink hybrid beamforming modules as an E2E neural network. While for FDD systems, we jointly model the downlink pilot transmission, uplink CSI feedback, and downlink hybrid beamforming modules as an E2E neural network. Different from conventional approaches separately processing different modules, the proposed solution simultaneously optimizes all modules with the sum rate as the optimization object. Therefore, by perceiving the inherent property of air-to-ground massive MIMO-OFDM channel samples, the DL-based E2E neural network can establish the mapping function from the channel to the beamformer, so that the explicit channel reconstruction can be avoided with reduced pilot and feedback overhead. Besides, practical low-resolution phase shifters (PSs) introduce the quantization constraint, leading to the intractable gradient backpropagation when training the neural network. To mitigate the performance loss caused by the phase quantization error, we adopt the transfer learning strategy to further fine-tune the E2E neural network based on a pre-trained network that assumes the ideal infinite-resolution PSs. Numerical results show that our DL-based schemes have considerable advantages over state-of-the-art schemes.

Improving the Reliability of Network Intrusion Detection Systems through Dataset Integration

This work presents Reliable-NIDS (R-NIDS), a novel methodology for Machine Learning (ML) based Network Intrusion Detection Systems (NIDSs) that allows ML models to work on integrated datasets, empowering the learning process with diverse information from different datasets. Therefore, R-NIDS targets the design of more robust models, that generalize better than traditional approaches. We also propose a new dataset, called UNK21. It is built from three of the most well-known network datasets (UGR'16, USNW-NB15 and NLS-KDD), each one gathered from its own network environment, with different features and classes, by using a data aggregation approach present in R-NIDS. Following R-NIDS, in this work we propose to build two well-known ML models (a linear and a non-linear one) based on the information of three of the most common datasets in the literature for NIDS evaluation, those integrated in UNK21. The results that the proposed methodology offers show how these two ML models trained as a NIDS solution could benefit from this approach, being able to generalize better when training on the newly proposed UNK21 dataset. Furthermore, these results are carefully analyzed with statistical tools that provide high confidence on our conclusions.

Active inference, Bayesian optimal design, and expected utility

Active inference, a corollary of the free energy principle, is a formal way of describing the behavior of certain kinds of random dynamical systems that have the appearance of sentience. In this chapter, we describe how active inference combines Bayesian decision theory and optimal Bayesian design principles under a single imperative to minimize expected free energy. It is this aspect of active inference that allows for the natural emergence of information-seeking behavior. When removing prior outcomes preferences from expected free energy, active inference reduces to optimal Bayesian design, i.e., information gain maximization. Conversely, active inference reduces to Bayesian decision theory in the absence of ambiguity and relative risk, i.e., expected utility maximization. Using these limiting cases, we illustrate how behaviors differ when agents select actions that optimize expected utility, expected information gain, and expected free energy. Our T-maze simulations show optimizing expected free energy produces goal-directed information-seeking behavior while optimizing expected utility induces purely exploitive behavior and maximizing information gain engenders intrinsically motivated behavior.

Mixed X-Ray Image Separation for Artworks with Concealed Designs

In this paper, we focus on X-ray images of paintings with concealed sub-surface designs (e.g., deriving from reuse of the painting support or revision of a composition by the artist), which include contributions from both the surface painting and the concealed features. In particular, we propose a self-supervised deep learning-based image separation approach that can be applied to the X-ray images from such paintings to separate them into two hypothetical X-ray images. One of these reconstructed images is related to the X-ray image of the concealed painting, while the second one contains only information related to the X-ray of the visible painting. The proposed separation network consists of two components: the analysis and the synthesis sub-networks. The analysis sub-network is based on learned coupled iterative shrinkage thresholding algorithms (LCISTA) designed using algorithm unrolling techniques, and the synthesis sub-network consists of several linear mappings. The learning algorithm operates in a totally self-supervised fashion without requiring a sample set that contains both the mixed X-ray images and the separated ones. The proposed method is demonstrated on a real painting with concealed content, Do\~na Isabel de Porcel by Francisco de Goya, to show its effectiveness.

Conditional Mutual information-based Contrastive Loss for Financial Time Series Forecasting

We present a method for financial time series forecasting using representation learning techniques. Recent progress on deep autoregressive models has shown their ability to capture long-term dependencies of the sequence data. However, the shortage of available financial data for training will make the deep models susceptible to the overfitting problem. In this paper, we propose a neural-network-powered conditional mutual information (CMI) estimator for learning representations for the forecasting task. Specifically, we first train an encoder to maximize the mutual information between the latent variables and the label information conditioned on the encoded observed variables. Then the features extracted from the trained encoder are used to learn a subsequent logistic regression model for predicting time series movements. Our proposed estimator transforms the CMI maximization problem to a classification problem whether two encoded representations are sampled from the same class or not. This is equivalent to perform pairwise comparisons of the training datapoints, and thus, improves the generalization ability of the deep autoregressive model. Empirical experiments indicate that our proposed method has the potential to advance the state-of-the-art performance.

Multi-Channel Multi-Speaker ASR Using 3D Spatial Feature

Automatic speech recognition (ASR) of multi-channel multi-speaker overlapped speech remains one of the most challenging tasks to the speech community. In this paper, we look into this challenge by utilizing the location information of target speakers in the 3D space for the first time. To explore the strength of proposed the 3D spatial feature, two paradigms are investigated. 1) a pipelined system with a multi-channel speech separation module followed by the state-of-the-art single-channel ASR module; 2) a "All-In-One" model where the 3D spatial feature is directly used as an input to ASR system without explicit separation modules. Both of them are fully differentiable and can be back-propagated end-to-end. We test them on simulated overlapped speech and real recordings. Experimental results show that 1) the proposed ALL-In-One model achieved a comparable error rate to the pipelined system while reducing the inference time by half; 2) the proposed 3D spatial feature significantly outperformed (31\% CERR) all previous works of using the 1D directional information in both paradigms.

Learning Physics through Images: An Application to Wind-Driven Spatial Patterns

For centuries, scientists have observed nature to understand the laws that govern the physical world. The traditional process of turning observations into physical understanding is slow. Imperfect models are constructed and tested to explain relationships in data. Powerful new algorithms are available that can enable computers to learn physics by observing images and videos. Inspired by this idea, instead of training machine learning models using physical quantities, we trained them using images, that is, pixel information. For this work, and as a proof of concept, the physics of interest are wind-driven spatial patterns. Examples of these phenomena include features in Aeolian dunes and the deposition of volcanic ash, wildfire smoke, and air pollution plumes. We assume that the spatial patterns were collected by an imaging device that records the magnitude of the logarithm of deposition as a red, green, blue (RGB) color image with channels containing values ranging from 0 to 255. In this paper, we explore deep convolutional neural network-based autoencoders to exploit relationships in wind-driven spatial patterns, which commonly occur in geosciences, and reduce their dimensionality. Reducing the data dimension size with an encoder allows us to train regression models linking geographic and meteorological scalar input quantities to the encoded space. Once this is achieved, full predictive spatial patterns are reconstructed using the decoder. We demonstrate this approach on images of spatial deposition from a pollution source, where the encoder compresses the dimensionality to 0.02% of the original size and the full predictive model performance on test data achieves an accuracy of 92%.