The Slodderwetenschap Sloppy Science of Stochastic Parrots A Plea for Science to NOT take the Route Advocated by Gebru and Bender

This article is a position paper written in reaction to the now-infamous paper titled “On the Dangers of Stochastic Parrots: Can Language Models Be TooBig?” by Timnit Gebru, Emily Bender, and others . I find the ethics of the Parrot Paper lacking, and in that lack, I worry about the direction in which computer science, machinelearning, and artificial intelligence are heading .…

The Slodderwetenschap Sloppy Science of Stochastic Parrots A Plea for Science to NOT take the Route Advocated by Gebru and Bender

This article is a position paper written in reaction to the now-infamous paper titled “On the Dangers of Stochastic Parrots: Can Language Models Be TooBig?” by Timnit Gebru, Emily Bender, and others . I find the ethics of the Parrot Paper lacking, and in that lack, I worry about the direction in which computer science, machinelearning, and artificial intelligence are heading .…

Neurocognitive Informatics Manifesto

Informatics studies all aspects of the structure of natural and artificial systems . Human information processing is still unmatched in many areas, including information management, representation and understanding . Neurocognitive informatics is a new, emerging field that should help to improve the matching of artificial and natural systems, and inspire better algorithms to solve problems that are still beyond the reach of machines .…

Target Detection and Segmentation in Circular Scan Synthetic Aperture Sonar Images using Semi Supervised Convolutional Encoder Decoders

We propose a saliency-based, multi-target detection and segmentation framework for multi-aspect, semi-coherent imagery formed from circular-scan,synthetic-aperture sonar (CSAS) Our framework relies on a multi-branch,convolutional encoder-decoder network (MB-CEDN) The encoder portion extract features from one or more CSAS images of the targets .…

Stabilized Nested Rollout Policy Adaptation

Nested Rollout Policy Adaptation (NRPA) is a Monte Carlo search algorithm for single player games . We propose to modify NRPA in order to improve the stability of the algorithm . Experiments show it improves the algorithm for different application domains: SameGame, Traveling Salesman withTime Windows and Expression Discovery .…

Summaformers LaySumm 20 LongSumm 20

Summarization is acognitively challenging task – extracting summary worthy sentences islaborious, and expressing semantics in brief is complicated . System is simple, intuitive and based on how specific paper sections contribute to human summaries . On blind test corpora, our system ranks first and third for the LongSumm and LaySumm tasks respectively.…

Entropic Causal Inference Identifiability and Finite Sample Results

Entropic causal inference is a framework for inferring the causal direction between two categorical variables from observational data . The central assumption is that the amount of unobserved randomness in the system is not toolarge . This is measured by the entropy of the exogenous variable in the underlying structural causal model, which governs the causalrelation between the observed variables .…

An Unsupervised Learning Method with Convolutional Auto Encoder for Vessel Trajectory Similarity Computation

The computation of vessel trajectorysimilarity has recently attracted increasing attention in the maritime datamining research community . Traditional shape- and warping-based methods often suffer from several drawbacks such as high computational cost and sensitivity to unwanted artifacts . Toeliminate these drawbacks, we propose an unsupervised learning method which extracts low-dimensional features through a convolutionalauto-encoder (CAE) The CAE can learn the low-dimension representationsof informative trajectory images in a manner .…

Provably Approximated ICP

The goal of the \emph{alignment problem is to align a point cloud $P= \{p_1,\cdots,p_n\}$ to another (observed) point cloud . Heuristics such as the IterativeClosest Point (ICP) algorithm and its variants were suggested for these problems . None yield a provable non-trivial approximation for the globaloptimum .…

Adaptive Prototypical Networks with Label Words and Joint Representation Learning for Few Shot Relation Classification

Relation classification (RC) task is one of fundamental tasks of informationextraction . We propose to encode each class prototype in an adaptive way from twoaspects . The results show that the proposed adaptive prototypical networks with label words and jointrepresentation learning has not only achieved significant inaccuracy, but also increased the generalization ability of few-shot RC models .…

BERT Family Eat Word Salad Experiments with Text Understanding

Ourexperiments show that state-of-the-art models consistently fail to recognizethem as ill-formed, and instead produce high confidence predictions on them . We define simple heuristics to construct such examples . We show that if models are explicitly trained to recognize invalid inputs, they can be robust to such attacks without a drop in performance .…

Quantum Secure Direct Communication with MutualAuthentication using a Single Basis

In this paper, we propose a new theoretical scheme for quantum secure direct communication (QSDC) with user authentication . Different from the previous QSDCprotocols, the present protocol uses only one orthogonal basis of single-qubitstates to encode the secret message . This is a one-time and one-way communication protocol, which uses qubits prepared in a randomly chosenarbitrary basis, to transmit the message .…

Kuksa Self Adaptive Microservices in Automotive Systems

Automotive systems should be self-adaptive, thereby they can make real-time decisions based on changing operating conditions . Emerging modern solutions, such as microservices could improveself-adaptation capabilities and ensure higher levels of quality performance in many domains . The results of our study indicate the importance of design trade-offs for qualityrequirements’ satisfaction levels of each microservices and the whole system for the optimal performance of an adaptive system at runtime.…

Beyond Helly graphs the diameter problem on absolute retracts

In this paper, we study the complexity ofcomputing the diameter within the absolute retracts of various hereditary graphclasses . We show how to compute the diameter of certain hereditary graphs in randomized time . For the special case of chordal bipartite graphs, it can be improved to linear time, and the algorithm even computes all the eccentricities .…

A Thermodynamic Core using Voltage Controlled Spin Orbit Torque Magnetic Tunnel Junctions

Magnetic devices within computing cores harness thermodynamics through its thermal stability . The evolution of network states isguided by the spin-orbit-torque effect . The results of this work pave the path towards the realization of high-performance thermodynamic computing hardware . Finally,this paper will also give a perspective of computing beyond thermodynamiccomputing.…

Using Crowdsourcing to Train Facial Emotion Machine Learning Models with Ambiguous Labels

Current emotion detection classifiers predict discrete emotions . Compound and ambiguous facialexpressions are often evoked by humans . Crowdsourced labels from the crowd tend to match the consensus from the originalCAFE raters, validating the utility of crowdsourcing . The output probability vector of the crowd-trained classifier much more closely resembles the distribution ofhuman labels (t=3.2827, p=0.0014) For many applications of affectivecomputing, reporting an emotion probability distribution that more closelyresembles human interpretation can be more important than traditional machinelearning metrics .…

Context Aware Target Apps Selection and Recommendation for Enhancing Personal Mobile Assistants

Users install many apps on their smartphones, raising issues related toinformation overload for users and resource management for devices . The recent increase in the use of personal assistants has made mobile devices even more pervasive in users’ lives . This paper addresses two research problemsthat are vital for developing effective personal mobile assistants: target appsselection and recommendation .…

Towards Long term Fairness in Recommendation

As Recommender Systems influence more and more people in their dailylife, the issue of fairness in recommendation is becoming more important . Most of the prior approaches to fairness-aware recommendation have been situated in a static or one-shot setting . This fails to consider the dynamic nature of the recommender systems, where attributes such as item popularity may changeover time due to the recommendation policy and user engagement .…

Randomised maximum likelihood based posterior sampling

Minimization of a stochastic cost function is commonly used for approximatesampling in high-dimensional Bayesian inverse problems with Gaussian priordistributions and multimodal posterior distributions . The density of the samples generated by minimization is not the desired target density, unless theobservation operator is linear, but the distribution of samples is useful as a proposal density for importance sampling or for Markov chain Monte Carlomethods .…

The shifted ODE method for underdamped Langevin MCMC

In this paper, we consider the underdamped Langevin diffusion (ULD) andpropose a numerical approximation using its associated ordinary differentialequation (ODE) When used as a Markov Chain Monte Carlo (MCMC) algorithm, the ODE approximation achieves a $2$-Wasserstein error of$\varepsilon$ in$\mathcal{O\big(d^{\frac{1)3) The main feature of the proposed numerical method isthat it can utilize additional smoothness of the target log-density $f$.…

Bandwidth Allocation for Multiple Federated Learning Services in Wireless Edge Networks

This paper studies a federated learning (FL) system where multiple services co-exist in a wireless network and share common wireless resources . It fills the void of wireless resource allocation for multiple simultaneous FL services in the existing literature . For cooperative FL service providers, we design a distributed bandwidth allocational algorithm to optimize the overall performance of multiple FL services, while cater to the fairness among FL services and the privacy of clients .…

Compliant Fins for Locomotion in Granular Media

In this paper, we present an approach to study the behavior of compliant plates in granular media and optimize the performance of a robot that utilizesthis technique for mobility . This concept utilizes one-sided joint limits to create anasymmetric gait cycle that avoids more complicated alternatives often found in other swimming/digging robots .…

Sub Goal Social Force Model for Collective Pedestrian Motion Under Vehicle Influence

Analytical modeling of collective pedestrian motion can benefit intelligent transportation practices like shared space design and urbanautonomous driving . In mixed traffic scenarios, a certain number of pedestrians might coexist in a small area while interacting with vehicles . In this situation, every pedestrian must simultaneously react to the surrounding pedestrians and vehicles .…

Cross Modal Contrastive Learning of Representations for Navigation using Lightweight Low Cost Millimeter Wave Radar for Adverse Environmental Conditions

Existing methods are limited by reliance on cameras and LiDAR devices, which have degraded sensing under adverse environmental conditions . In response, we propose the use of single-chipmillimeter-wave (mmWave) radar, which is lightweight and inexpensive, for learning-based autonomous navigation . Our proposed end-to-end deep RLpolicy with contrastive learning successfully navigated the robot through smoke-filled maze environments and achieved better performance compared withgenerative reconstruction methods .…

Reinforcement Learning Enabled Automatic Impedance Control of a Robotic Knee Prosthesis to Mimic the Intact Knee Motion in a Co Adapting Environment

Automatically configuring a robotic prosthesis to fit its user’s needs and conditions is a great technical challenge and a roadblock to the adoption of the technology . Previously, we have successfully developedreinforcement learning (RL) solutions toward addressing this issue . Yet, ourdesigns were based on using a subjectively prescribed target motion profile for the robotic knee during level ground walking .…

Learning Student Interest Trajectory for MOOCThread Recommendation

Massive Open Online Courses (MOOCs) have witnessed immensegrowth in popularity . Discussion forums are primarymeans of interaction among learners and instructors . With growingclass size, students face the challenge of finding useful and informativediscussion forums . The fundamental challenge is that the studentinterests drift as they progress through the course, and forum contents evolveas students or instructors update them .…

Locating Faults with Program Slicing An Empirical Analysis

Study of 457 bugs (369 single faults and 88 multiple faults) in 46 C programs . Dynamic slicing is more effective for programs with single faults, but statistical debugging performs better on multiple faults . Best results obtained by a hybrid approach: If programmers first examine atmost the top five most suspicious locations from statistical debugging, and then switch to dynamic slices, on average, they will need to examine 15% (30 lines) of the code .…