While Generative Adversarial Networks (GANs) achieve spectacular results on unstructured data like images, there is still a gap on tabular data, data for which state of the art supervised learning still favours to a large extent decision tree (DT)-based models. This paper proposes a new path forward for the generation of tabular data, exploiting decades-old understanding of the supervised task's best components for DT induction, from losses (properness), models (tree-based) to algorithms (boosting). The \textit{properness} condition on the supervised loss -- which postulates the optimality of Bayes rule -- leads us to a variational GAN-style loss formulation which is \textit{tight} when discriminators meet a calibration property trivially satisfied by DTs, and, under common assumptions about the supervised loss, yields "one loss to train against them all" for the generator: the $\chi^2$. We then introduce tree-based generative models, \textit{generative trees} (GTs), meant to mirror on the generative side the good properties of DTs for classifying tabular data, with a boosting-compliant \textit{adversarial} training algorithm for GTs. We also introduce \textit{copycat training}, in which the generator copies at run time the underlying tree (graph) of the discriminator DT and completes it for the hardest discriminative task, with boosting compliant convergence. We test our algorithms on tasks including fake/real distinction, training from fake data and missing data imputation. Each one of these tasks displays that GTs can provide comparatively simple -- and interpretable -- contenders to sophisticated state of the art methods for data generation (using neural network models) or missing data imputation (relying on multiple imputation by chained equations with complex tree-based modeling).
This paper presents a novel channel estimation technique for the multi-user massive multiple-input multiple-output (MU-mMIMO) systems using angular-based hybrid precoding (AB-HP). The proposed channel estimation technique generates group-wise channel state information (CSI) of user terminal (UT) zones in the service area by deep neural networks (DNN) and fuzzy c-Means (FCM) clustering. The slow time-varying CSI between the base station (BS) and feasible UT locations in the service area is calculated from the geospatial data by offline ray tracing and a DNN-based path estimation model associated with the 1-dimensional convolutional neural network (1D-CNN) and regression tree ensembles. Then, the UT-level CSI of all feasible locations is grouped into clusters by a proposed FCM clustering. Finally, the service area is divided into a number of non-overlapping UT zones. Each UT zone is characterized by a corresponding set of clusters named as UT-group CSI, which is utilized in the analog RF beamformer design of AB-HP to reduce the required large online CSI overhead in the MU-mMIMO systems. Then, the reduced-size online CSI is employed in the baseband (BB) precoder of AB-HP. Simulations are conducted in the indoor scenario at 28 GHz and tested in an AB-HP MU-mMIMO system with a uniform rectangular array (URA) having 16x16=256 antennas and 22 RF chains. Illustrative results indicate that 91.4% online CSI can be reduced by using the proposed offline channel estimation technique as compared to the conventional online channel sounding. The proposed DNN-based path estimation technique produces same amount of UT-level CSI with runtime reduced by 65.8% as compared to the computationally expensive ray tracing.
It is well-known that the Burer-Monteiro (B-M) factorization approach can efficiently solve low-rank matrix optimization problems under the RIP condition. It is natural to ask whether B-M factorization-based methods can succeed on any low-rank matrix optimization problems with a low information-theoretic complexity, i.e., polynomial-time solvable problems that have a unique solution. In this work, we provide a negative answer to the above question. We investigate the landscape of B-M factorized polynomial-time solvable matrix completion (MC) problems, which are the most popular subclass of low-rank matrix optimization problems without the RIP condition. We construct an instance of polynomial-time solvable MC problems with exponentially many spurious local minima, which leads to the failure of most gradient-based methods. Based on those results, we define a new complexity metric that potentially measures the solvability of low-rank matrix optimization problems based on the B-M factorization approach. In addition, we show that more measurements of the ground truth matrix can deteriorate the landscape, which further reveals the unfavorable behavior of the B-M factorization on general low-rank matrix optimization problems.
The Intelligent decision of the unmanned combat aerial vehicle (UCAV) has long been a challenging problem. The conventional search method can hardly satisfy the real-time demand during high dynamics air combat scenarios. The reinforcement learning (RL) method can significantly shorten the decision time via using neural networks. However, the sparse reward problem limits its convergence speed and the artificial prior experience reward can easily deviate its optimal convergent direction of the original task, which raises great difficulties for the RL air combat application. In this paper, we propose a homotopy-based soft actor-critic method (HSAC) which focuses on addressing these problems via following the homotopy path between the original task with sparse reward and the auxiliary task with artificial prior experience reward. The convergence and the feasibility of this method are also proved in this paper. To confirm our method feasibly, we construct a detailed 3D air combat simulation environment for the RL-based methods training firstly, and we implement our method in both the attack horizontal flight UCAV task and the self-play confrontation task. Experimental results show that our method performs better than the methods only utilizing the sparse reward or the artificial prior experience reward. The agent trained by our method can reach more than 98.3% win rate in the attack horizontal flight UCAV task and average 67.4% win rate when confronted with the agents trained by the other two methods.
The study forms a technical report of various tasks that have been performed on the materials collected and published by Finnish ethnographer and linguist, Matthias Alexander Castr\'en (1813-1852). The Finno-Ugrian Society is publishing Castr\'en's manuscripts as new critical and digital editions, and at the same time different research groups have also paid attention to these materials. We discuss the workflows and technical infrastructure used, and consider how datasets that benefit different computational tasks could be created to further improve the usability of these materials, and also to aid the further processing of similar archived collections. We specifically focus on the parts of the collections that are processed in a way that improves their usability in more technical applications, complementing the earlier work on the cultural and linguistic aspects of these materials. Most of these datasets are openly available in Zenodo. The study points to specific areas where further research is needed, and provides benchmarks for text recognition tasks.
The rise of deep learning algorithms has led many researchers to withdraw from using classic signal processing methods for sound generation. Deep learning models have achieved expressive voice synthesis, realistic sound textures, and musical notes from virtual instruments. However, the most suitable deep learning architecture is still under investigation. The choice of architecture is tightly coupled to the audio representations. A sound's original waveform can be too dense and rich for deep learning models to deal with efficiently - and complexity increases training time and computational cost. Also, it does not represent sound in the manner in which it is perceived. Therefore, in many cases, the raw audio has been transformed into a compressed and more meaningful form using upsampling, feature-extraction, or even by adopting a higher level illustration of the waveform. Furthermore, conditional on the form chosen, additional conditioning representations, different model architectures, and numerous metrics for evaluating the reconstructed sound have been investigated. This paper provides an overview of audio representations applied to sound synthesis using deep learning. Additionally, it presents the most significant methods for developing and evaluating a sound synthesis architecture using deep learning models, always depending on the audio representation.
We present a polynomial-time algorithm that discovers all maximal patterns in a point set, $D\in\mathbb{R}^k$, that are related by transformations in a user-specified class, $F$, of bijections over $\mathbb{R}^k$. We also present a second algorithm that discovers the set of occurrences for each of these maximal patterns and then uses compact encodings of these occurrence sets to compute a losslessly compressed encoding of the input point set. This encoding takes the form of a set of pairs, $E=\left\lbrace\left\langle P_1, T_1\right\rangle,\left\langle P_2, T_2\right\rangle,\ldots\left\langle P_{\ell}, T_{\ell}\right\rangle\right\rbrace$, where each $\langle P_i,T_i\rangle$ consists of a maximal pattern, $P_i\subseteq D$, and a set, $T_i\subset F$, of transformations that map $P_i$ onto other subsets of $D$. Each transformation is encoded by a vector of real values that uniquely identifies it within $F$ and the length of this vector is used as a measure of the complexity of $F$. We evaluate the new compression algorithm with three transformation classes of differing complexity, on the task of classifying folk-song melodies into tune families. The most complex of the classes tested includes all combinations of the musical transformations of transposition, inversion, retrograde, augmentation and diminution. We found that broadening the transformation class improved performance on this task. However, it did not, on average, improve compression factor, which may be due to the datasets (in this case, folk-song melodies) being too short and simple to benefit from the potentially greater number of pattern relationships that are discoverable with larger transformation classes.
There exists a distribution discrepancy between training and testing, in the way images are fed to modern CNNs. Recent work tried to bridge this gap either by fine-tuning or re-training the network at different resolutions. However re-training a network is rarely cheap and not always viable. To this end, we propose a simple solution to address the train-test distributional shift and enhance the performance of pre-trained models -- which commonly ship as a package with deep learning platforms \eg, PyTorch. Specifically, we demonstrate that running inference on the center crop of an image is not always the best as important discriminatory information may be cropped-off. Instead we propose to combine results for multiple random crops for a test image. This not only matches the train time augmentation but also provides the full coverage of the input image. We explore combining representation of random crops through averaging at different levels \ie, deep feature level, logit level, and softmax level. We demonstrate that, for various families of modern deep networks, such averaging results in better validation accuracy compared to using a single central crop per image. The softmax averaging results in the best performance for various pre-trained networks without requiring any re-training or fine-tuning whatsoever. On modern GPUs with batch processing, the paper's approach to inference of pre-trained networks, is essentially free as all images in a batch can all be processed at once.
By using a computer keyboard as a finger recording device, we construct the largest existing dataset for gesture recognition via surface electromyography (sEMG), and use deep learning to achieve over 90% character-level accuracy on reconstructing typed text entirely from measured muscle potentials. We prioritize the temporal structure of the EMG signal instead of the spatial structure of the electrode layout, using network architectures inspired by those used for real-time spoken language transcription. Our architecture recognizes the rapid movements of natural computer typing, which occur at irregular intervals and often overlap in time. The extensive size of our dataset also allows us to study gesture recognition after synthetically downgrading the spatial or temporal resolution, showing the system capabilities necessary for real-time gesture recognition.
Taking advantage of both vehicle-to-everything (V2X) communication and automated driving technology, connected and automated vehicles are quickly becoming one of the transformative solutions to many transportation problems. However, in a mixed traffic environment at signalized intersections, it is still a challenging task to improve overall throughput and energy efficiency considering the complexity and uncertainty in the traffic system. In this study, we proposed a hybrid reinforcement learning (HRL) framework which combines the rule-based strategy and the deep reinforcement learning (deep RL) to support connected eco-driving at signalized intersections in mixed traffic. Vision-perceptive methods are integrated with vehicle-to-infrastructure (V2I) communications to achieve higher mobility and energy efficiency in mixed connected traffic. The HRL framework has three components: a rule-based driving manager that operates the collaboration between the rule-based policies and the RL policy; a multi-stream neural network that extracts the hidden features of vision and V2I information; and a deep RL-based policy network that generate both longitudinal and lateral eco-driving actions. In order to evaluate our approach, we developed a Unity-based simulator and designed a mixed-traffic intersection scenario. Moreover, several baselines were implemented to compare with our new design, and numerical experiments were conducted to test the performance of the HRL model. The experiments show that our HRL method can reduce energy consumption by 12.70% and save 11.75% travel time when compared with a state-of-the-art model-based Eco-Driving approach.