Exponential Savings in Agnostic Active Learning through Abstention

We show that in pool-based active classification without assumptions on the underlying distribution, if the learner is given the power to abstain from somepredictions by paying the price marginally smaller than the average loss $1/2$ of a random guess, exponential savings in the number of label requests are possible whenever they are possible in the corresponding realizable problem .…

Learning elliptic partial differential equations with randomized linear algebra

Given input-output pairs of an elliptic partial differential equation (PDE) in three dimensions, we derive the first theoretically-rigorous scheme for learning the associated Green’s function $G$ The quantity $0<\epsilon<1$characterizes the quality of the training dataset . Along the way, we extend therandomized singular value decomposition algorithm for learning matrices toHilbert--Schmidt operators and characterize quality of covariance kernels for PDE learning . …

Computation of multi degree Tchebycheffian B splines

Multi-degree Tchebycheffian splines are splines with pieces drawn fromextended (complete) Tche bycheff spaces . These are a natural extension of multi-degree polynomial splines . Wepresent a practical framework to compute MDTB-splines, and provide anobject-oriented implementation in Matlab . The implementation supports the construction, differentiation, and visualization of MDTB splines whose piecesbelong to TcheByCheff spaces that are null-spaces of constant-coefficientlinear differential operators .…

Solving the linear semiclassical Schrödinger equation on the real line

The numerical solution of a linear Schr\”odinger equation in thesemiclassical regime is very well understood in a torus $\mathbb{T}^d$ A raft of modern computational methods are precise and affordable, while conserving energy and resolving high oscillations very well . This is far from the case with regard to its solution in a setting more suitable for many applications .…

A parallel in time two sided preconditioning for all at once system from a non local evolutionary equation with weakly singular kernel

In this paper, we study a parallel-in-time (PinT) algorithm for all-at-oncesystem from a non-local evolutionary equation with weakly singular kernel . We propose to use a two-sided preconditioning technique for theall at-once discretization of the equation . This is the first attempt to develop a PinTpreconditioning technique that has fast and exact implementation and that the .responding…

Demonstrating the Evolution of GANs through t SNE

Generative Adversarial Networks (GANs) are powerful generative models thatachieved strong results, mainly in the image domain . Evolutionary algorithms, such as COEGAN, were recently proposed as a solution to improve the GAN training . We propose an evaluation method based on t-distributed Stochastic NeighbourEmbedding (t-SNE) to assess the progress of GANs and visualize the distributionlearned by generators in training .…

LPWAN in the TV White Spaces A Practical Implementation and Deployment Experiences

Low-Power Wide-Area Network (LPWAN) is an enabling Internet-of-Things (IoT) technology that supports long-range, low-power, and low-cost connectivity ton numerous devices . SNOW (Sensor Networkover White Spaces) is a promising LPWAN platform that operates over the TVwhite spaces . The current SNOW implementation uses the USRP devices as LPWan nodes, which has high costs (~$750 USD per device) and large form-factors, hindering its applicability in practical deployment .…

Model Based Testing of Networked Applications

We present a rigorous framework for automatically testing application-layerprotocols . Key innovation is a domain-specific embedded language for writing nondeterministic models of the behavior of networked servers . We demonstrate effectiveness of this framework by using it to specify andtest a fragment of HTTP/1.1 .…

Using Bayesian Modelling to Predict Software Incidents

Fault- or event-tree analyses have been used to estimate the probability of a safety-critical device creating a dangerous condition . However, these analysis techniques are less effective for systems reliant on software, and are perhaps least effective in Safety of theIntended Functionality (SOTIF) environments .…

Lyapunov Based Stabilization and Control of Closed Quantum Systems

A Lyapunov-based method is presented for stabilizing and controlling of closed quantum systems . The proposed method can potentially be applied for high-fidelity quantum control purposes in quantum computing frameworks . Results show significant improvement in both the set of stabilizable quantumsystems and their invariant sets of state trajectories generated by designedcontrol signals .…

Retro Reflective Beam Communications with Spatially Separated Laser Resonator

Resonant beamcommunications (RBCom) is an OWC technology which satisfies theneeds of non-mechanical mobility and high signal-to-noise ratio (SNR) It hasthe self-alignment feature and therefore avoids positioning and pointingoperations . However, RBCom undergoes echo interference . The transmitter and the receiver constitute a spatially separatedlaser resonator, in which the retro-reflective resonant beam is formed andtracks the receiver automatically .…

A Novel Use of Discrete Wavelet Transform Features in the Prediction of Epileptic Seizures from EEG Data

This paper demonstrates the predictive superiority of discrete wavelettransform (DWT) over previously used methods of feature extraction in thediagnosis of epileptic seizures from EEG data . The mean-differences are statistically significant respectively in the imbalanced and balanced dataset . The results also highlight that MFCC performs less than all the DWT used in this study and that, MFCC does less than .…

Bandgap optimization in combinatorial graphs with tailored ground states Application in Quantum annealing

A mixed-integer linear programming (MILP) formulation is presented forparameter estimation of the Potts model . This is useful in the development of energy-based graph models to be simulated on Quantum annealing hardware where the exact simulationtemperature is unknown . Computationally, the memory requirement in this method grows exponentially with the graph size.…

Community Detection Exact Recovery in Weighted Graphs

In community detection, the exact recovery of communities (clusters) has been investigated under the general stochastic block model with edges drawn from Bernoulli distributions . We introduce a new semi-metric that describes sufficient and necessary conditions of exact recovery . The necessary and sufficient conditions are areasymptotically tight .…

Speech Recognition by Simply Fine tuning BERT

We propose a simple method for automatic speech recognition (ASR) by fine-tuning BERT . BERT is a language model (LM) trained on large-scaleunlabeled text data and can generate rich contextual representations . Our assumption is that given a history context sequence, a powerful LM can narrow the range of possible choices and the speech signal can be used as a simpleclue .…

Fake it Till You Make it Self Supervised Semantic Shifts for Monolingual Word Embedding Tasks

The use of language is subject to variation over time as well as across social groups and knowledge domains, leading to differences even in themonolingual scenario . Such variation in word usage is often called lexicalsemantic change (LSC) The goal of LSC is to characterize and quantify languagevariations with respect to word meaning, to measure how distinct two languagesources are .…

Machine Translationese Effects of Algorithmic Bias on Linguistic Complexity in Machine Translation

Machine Translation (MT) and Natural LanguageProcessing (NLP) have shown that existing models amplify biases observed in the training data . We hypothesize that the’algorithmic bias’, i.e. an exacerbation of frequently observed patterns incombination with a loss of less frequent ones, not only exacerbates societalbiases present in current datasets but could also lead to an artificiallyimpoverished language: ‘machine translationese’ We assess the linguisticrichness (on a lexical and morphological level) of translations created by different data-driven MT paradigms – phrase-based statistical (PB-SMT) andneural MT (NMT) Our experiments show that there is a .…

Taxonomic survey of Hindi Language NLP systems

Natural Language processing (NLP) represents the task of automatic handling of natural human language by machines . Hindi is the official language of India with nearly 691 million users in India and 366 million in rest of world . There is large spectrum of possible applications of NLP which help in automating tasks like translating text from one language to other, retrieving and summarizing data from very hugerepositories, spam email filtering, identifying fake news in digital media, find political opinions and views of people on various government policies, provide effective medical assistance based on past history records of patient etc.…

Can We Automate Scientific Reviewing

The rapid development of science and technology has been accompanied by anexponential growth in peer-reviewed scientific publications . The review of each paper is a laborious process that must be carried out by subject matter experts . We discuss the possibility of using state-of-the-art natural language processing (NLP) models to generate first-pass peer reviews for scientific papers .…

Efficient CNN Building Blocks for Encrypted Data

Machine learning on encrypted data can address concerns related toprivacy and legality of sharing sensitive data with untrustworthy service providers . Fully Homomorphic Encryption (FHE) is a promising technique toenable machine learning and inferencing while providing strict guarantees against information leakage .…

Zur Integration von Post Quantum Verfahren in bestehende Softwareprodukte

PQC algorithms are being standardized to address the emerging threat to conventional asymmetric algorithms from quantum computing . These new algorithms must then be integrated into existing protocols, applications and infrastructures . Integration problems are to be expected, due toincompatibilities with existing standards and implementations on the one hand, but also due to a lack of knowledge among software developers about how to handle the algorithms .…