We address the problem of learning an efficient and adaptive physical layer encoding to communicate binary information over impaired channel. In contrast traditional work, we treat unsupervised machine focusing on optimizing reconstruction loss through artificial impairment layers in autoencoder (we term this a channel autoencoder) introduce several new regularizing which emulate common wireless impairments. also discuss role attention models form radio transformer network for helping...

Estimation is a critical component of synchronization in wireless and signal processing systems. There rich body work on estimator derivation, optimization, statistical characterization from analytic system models which are used pervasively today. We explore an alternative approach to building estimators relies principally approximate regression using large datasets computationally efficient artificial neural network capable learning non-linear function mappings provide compact accurate...

A modern power system is characterized by an increasing penetration of wind power, which results in large uncertainties its states. These must be quantified properly; otherwise, the security may threatened. Facing this challenge, we propose a cost-effective, data-driven approach to assessing system's load margin probabilistically. Using actual data, kernel density estimator applied infer nonparametric speed distributions, are further merged into framework vine copula. The latter enables us...

In this paper, we propose a new data poisoning attack and apply it to deep reinforcement learning agents. Our centers on what call in-distribution triggers, which are triggers native the distributions model will be trained deployed in. We outline simple procedure for embedding these, other, in agents following multi-task paradigm, demonstrate three common environments. believe that work has important implications security of models.

We use multiple measures of graph complexity to evaluate the realism synthetically-generated networks human activity, in comparison with several stylized network models as well a collection empirical from literature. The synthetic are generated by integrating data about populations sources, including Census, transportation surveys, and geographical data. resulting represent an approximation daily or weekly interaction. Our results indicate that synthetically graphs according our methodology...

We present a crop simulation environment with an OpenAI Gym interface, and apply modern deep reinforcement learning (DRL) algorithms to optimize yield. empirically show that DRL may be useful in discovering new policies approaches help yield, while simultaneously minimizing constraining factors such as water fertilizer usage. propose this hybrid plant modeling data-driven approach for strategies yield address upcoming global food demands due population expansion climate change.

Feature selection in machine learning aims to find out the best subset of variables from input that reduces computation requirement and improves predictor performance. In this paper, a new index based on empirical copulas, termed Copula Statistic (CoS) assess strength statistical dependence for testing independence is introduced. It shown test exhibits higher power than other indices. Finally, CoS applied feature problems, which allow demonstration good performance CoS.

A new index based on empirical copulas, termed the Copula Statistic (CoS), is introduced for assessing strength of multivariate dependence and testing statistical independence. New properties copulas are proved. They allow us to define CoS in terms a relative distance function between copula, Fr\'echet-Hoeffding bounds independence copula. Monte Carlo simulations reveal that large sample sizes, approximately normal. This property utilised develop CoS-based test against various noisy...

The dynamics of a power system with significant presence renewable energy resources are growing increasingly nonlinear. This nonlinearity is result the intermittent nature these and switching behavior their electronic devices. Therefore, it crucial to address in blind source separation methods. In this paper, we propose linear mixture dependent sources based on copula statistics that measure non-linear dependence between component signals structured as density functions. assumed be...

The increasing penetration of renewable energy along with the variations loads bring large uncertainties in power system states that are threatening security planning and operation. Facing these challenges, this paper proposes a cost-effective, nonparametric method to quantity impact uncertain injections on load margins. First, we propose generate inputs via novel vine copula due its capability simulating complex multivariate highly dependent model inputs. Furthermore, reduce prohibitive...

In this paper, we introduce the TrojAI software framework, an open source set of Python tools capable generating triggered (poisoned) datasets and associated deep learning (DL) models with trojans at scale. We utilize developed framework to generate a large trojaned MNIST classifiers, as well demonstrate capability produce reinforcement-learning model using vector observations. Results on show that nature trigger, training batch size, dataset poisoning percentage all affect successful...

Many modern systems for speaker diarization, such as the recently-developed VBx approach, rely on clustering of DNN embeddings followed by resegmentation.Two problems with this approach are that is not directly optimized task, and parameters need significant retuning different applications.We have recently presented progress in direction a Leave-One-Out Gaussian PLDA (LGP) algorithm an to training optimize performance scoring method.This paper presents new two-pass version system, where...

This paper presents a modern method for implementing burst modems in GNU Radio. Since are widely used multi-user channel access and sharing non-broadcast radio systems, this capability is critical to the development of numerous waveforms We focus on making such systems easy develop adapt wide classes computationally efficient at runtime. use Radio Event Stream scheduler demonstrate concise implementations PSK FSK compare with alternate approaches which have been attempted

We address the problem of learning efficient and adaptive ways to communicate binary information over an impaired channel. treat as reconstruction optimization through impairment layers in a channel autoencoder introduce several new domain-specific regularizing emulate common impairments. also apply radio transformer network based attention model on input decoder help recover canonical signal representations. demonstrate some promising initial capacity results from this architecture...

In this paper, we outline a framework for modeling utility-based blockchain-enabled economic systems using Agent Based Modeling (ABM). Our approach is to model the supply dynamics based on metrics of cryptoeconomy. We then build autonomous agents that make decisions those metrics. Those decisions, in turn, impact next time-step, creating closed loop models evolution cryptoeconomies over time. apply as case-study Filecoin, decentralized blockchain-based storage network. perform several...

The rewards a blockchain miner earns vary with time. Most of the time is spent mining without receiving any rewards, and only occasionally wins block reward. Mining pools smoothen stochastic flow in ideal case, provide steady over Smooth allow miners to choose an optimal power growth strategy that will result higher reward yield for given investment. We quantify economic advantage having smooth use this define maximum percentage should be willing pay pool services.

The dynamics of a power system with significant presence renewable energy resources are growing increasingly nonlinear. This nonlinearity is result the intermittent nature these and switching behavior their electronic devices. Therefore, it crucial to address in blind source separation methods. In this paper, we propose linear mixture dependent sources based on copula statistics that measure non-linear dependence between component signals structured as density functions. assumed be...