## Slipping to the Extreme A Mixed Method to Explain How Extreme Opinions Infiltrate Online Discussions

Aims to accelerate qualitative analysis of problematic onlinespeech by leveraging machine learning algorithms . Authors present case studies of the dynamics of problematicspeech in a far-right Facebook group, exemplifying its mutation from conservative to extreme . Using the latter, they examine dynamics of opinionemergence and co-occurrence, and we hint at some of the pathways through whichextreme opinions creep into the mainstream online discourse.…

## MWPToolkit An Open Source Framework for Deep Learning Based Math Word Problem Solvers

Developing automatic Math Word Problem (MWP) solvers has been an interest of NLP researchers since the 1960s . InMWPToolkit, we decompose the procedure of existing MWP solvers into multiplecore components and decouple their models into highly reusable modules . We alsoprovide a hyper-parameter search function to boost the performance .…

## Towards disease aware image editing of chest X rays

Disease-aware image editing by means of generative adversarial networks constitutes a promising avenue for advancing the use of AI in the healthcare sector . This work was presented in the MedicalImaging meets Neurips Workshop 2020, which was held as part of the 34thConference on Neural Information Processing Systems (NeurIPS 2020) in Vancouver, Canada .…

## So Cloze yet so Far N400 Amplitude is Better Predicted by Distributional Information than Human Predictability Judgements

More predictable words are easier to process – they are read faster and . smaller neural signals associated with processing difficulty, mostnotably, the N400 component of the event-related brain potential . Study finds that the predictions of three top-of-the-line . contemporary language models match the .…

## Matrix oriented FEM formulation for stationary and time dependent PDEs on x normal domains

Matrix-Orientedformulation of the classical Finite Element Method, called MO-FEM, of arbitrary order $k\in\mathbb{N}$. On structured 2D domains (e.g. squares or rectangles) the discrete problem is then reformulated as a Sylvester matrix equation . We apply the matrix-orientedform of the Preconditioned Conjugate Gradient (MO-PCG) method to solve each multiterm Syvester equation for MO-fEM of degree $k=1,\dots,4$ and for the lumped case .…

## A Physics Data Driven Bayesian Method for Heat Conduction Problems

A novel physics-data-driven Bayesian method named HeatConduction Equation assisted Bayesian Neural Network is proposed . The HCE-BNN is constructed based on the Bayesian neural network, it is aphysics-informed machine learning strategy . Compared with the existed pure datadriven method, to acquire physical consistency and better performance of the data-driven model, the heat conduction equation is embedded into the lossfunction of the HCE -BNN as a regularization term .…

## GNSS Outlier Mitigation Via Graduated Non Convexity Factor Graph Optimization

Accurate and globally referenced global navigation satellite system (GNSS)based vehicular positioning can be achieved in outlier-free open areas . Performance of GNSS can be significantly degraded by outlier measurements, such as multipath effects and non-line-of-sight (NLOS) receptionsarising from signal reflections of buildings .…

## Establishing Interlingua in Multilingual Language Models

Large multilingual language models show remarkable zero-shot cross-lingualtransfer performance on a range of tasks . Follow-up works hypothesized that these models internally project representations of different languages into ashared interlingual space . In this paper, we correct %one of the previous works the famous prior workclaiming that “BERT is not an Interlingua” and show that with the proper choice of sentence representation different languages actually do converge to a shared space in such language models .…

## Benchmarking the Robustness of Instance Segmentation Models

This paper presents a comprehensive evaluation of instance segmentation models with respect to real-world image corruptions . The out-of-domain image evaluation shows thegeneralization capability of models . The study includes state-of theart networkarchitectures, network backbones, normalization layers, models trained startingfrom scratch or ImageNet pretrained networks, and the effect of multi-tasktraining on robustness and generalization on the models .…

## Coordinating Narratives and the Capitol Riots on Parler

Ageneral methodology to uncover coordinated messaging through analysis of userparleys on Parler . We study three influential groups of users in the 6 January 2020Capitol riots and detect networks of coordinated user clusters that are allposting similar textual content in support of different disinformationnarratives related to the U.S.…

## Energy Efficient Multi Orchestrator Mobile Edge Learning

Mobile Edge Learning (MEL) is a collaborative learning paradigm that features distributed training of Machine Learning (ML) models over edge devices (e.g.,IoT devices) In MEL, possible coexistence of multiple learning tasks may arise . Multiobjective optimization problem is formulated to minimize the total energyconsumption and maximize the learning tasks’ accuracy .…

## MitoDet Simple and robust mitosis detection

Mitotic figure detection is a challenging task in digital pathology that has a direct impact on therapeutic decisions . The problem can be mainly attributed to aphenomenon called domain shift . An important source of a domain shift isintroduced by different microscopes and their camera systems, which noticeably change the color representation of digitized images .…

## Solving the Discrete Euler Arnold Equations for the Generalized Rigid Body Motion

The Moser-Veselov equation arises in the discretization of the Euler-Arnold differential equations governing the motion of a generalized rigid body . We propose three iterative methods for solving the Moser Veselov equations . The first onesplits the orthogonal constraints using the Bregman method .…

## Domain Robust Mitotic Figure Detection with StyleGAN

We propose a new training scheme for domain generalization in mitotic figuredetection . By considering the image variance due to different scanner types as different image styles, we have trained our detection network to be robust on scanner types . The proposed style transfer module generates different styled images from an input image with random code, eventuallygenerating variously styled images .…

## Do Prompt Based Models Really Understand the Meaning of their Prompts

A boom of papers have shown extraordinary progress in few-shotlearning with various prompt-based models . We find that models learn just as fast with many prompts that are intentionally irrelevant or even pathologically misleading as they do with instructively “good” prompts .…

## Optimal Target Shape for LiDAR Pose Estimation

Targets are essential in problems such as object tracking in cluttered environments, camera (and multi-sensor) calibration tasks, and SLAM . Symmetric shapes lead to pose ambiguity when using sparse sensor data such as LiDAR point clouds . A target is designed to induce largegradients at edge points under rotation to ameliorate the quantization uncertainty associated with point cloudsparseness .…

## Universal and Tight Online Algorithms for Generalized Mean Welfare

We study fair and efficient allocation of divisible goods, in an onlinemanner, among $n$ agents . The goods arrive online in a sequence of $T$ time periods . The agents’ values for a good are revealed only after its arrival, and the algorithm needs to fractionally allocate the good, immediately andirrevocably, among the agents .…

Roadscene2vec is an open-source tool for extracting and embedding road scene-graphs . The tool is designed to enable research into applications and capabilities of road scenes . It is available athttps://://github.com/AICPS/roadscene-2vec and is available in the wild at http://www.cicPS.org/roadscenes2vec . The capabilities of the tool include customized scenes-graph generation from either video clips or data from theCARLA simulator, built-in functionality for using graph and sequence embeddings for risk assessment and collision prediction applications, and utilities for visualizing and analyzing the explainability of graph learning models .…

## Construction of Inter Group Complementary Code Set and 2 D Z Complementary Array Code Set Based on Multivariable Functions

The need of two-D arrays with good 2-D correlation properties has been found of great interest in research because of the rapid development of the modern technologies in the field of wireless communications . The use ofmultivariable function enables us to set the size of the ZCACS to be $L_1\times L_2$ and $m_\alpha \geq 2$ for all $\alpha=1,2, \dots, k$, which is animprovement over the traditional generalized Boolean function that producessequences of length of the form $2^m$.…

## An Empirical Exploration in Quality Filtering of Text Data

We find aggressive filtering can in fact lead to adecrease in model quality on a wide array of downstream tasks for a GPT-like language model . We speculate that optimizing sufficientlystrongly for a proxy metric harms performance on the true objective .…

## Solving Inverse Problems with Conditional GAN Prior via Fast Network Projected Gradient Descent

Anetwork-based projected gradient descent (NPGD) algorithm formeasurement-conditional generative models to solve the inverse problem muchfaster than regular PGD . The achieved reconstruction speed-up in ourexperiments is up to 140-175. We combine the NPGD with conditional GAN/BEGAN toevaluate their effectiveness in solving compressed sensing type problems .…

## So Cloze yet so Far N400 Amplitude is Better Predicted by Distributional Information than Human Predictability Judgements

More predictable words are easier to process – they are read faster and . smaller neural signals associated with processing difficulty, mostnotably, the N400 component of the event-related brain potential . Study finds that the predictions of three top-of-the-line . contemporary language models match the .…

## GPU accelerated Optimal Path Planning in Stochastic Dynamic Environments

Autonomous marine vehicles play an essential role in many ocean science and engineering applications . Planning time and energy optimal paths for these vehicles to navigate in stochastic dynamic ocean environments is essential to reduce operational costs . Building a realistic model and solving the modeled MDPbecomes computationally expensive in large-scale real-time applications .…

## TrouSPI Net Spatio temporal attention on parallel atrous convolutions and U GRUs for skeletal pedestrian crossing prediction

TrouSPI-Net is a context-free, lightweight, multi-branchpredictor . It links the dynamics of a pedestrian’s skeleton to a binary crossingintention . The proposed approach is then enhanced by processing features such as relativedistances of skeletal joints, bounding box positions, or ego-vehicle speed with U-GRUs .…

## Supporting CUDA for an extended RISC V GPU architecture

RISC-V is the most popular choice for hardware ISA, thanks to itselegant design and open-source license . We have succeeded in executing CUDA kernels with several features, like multi-thread and atomic instructions, on an Risc-V GPUarchitecture . In this project, we aim to utilizethesesese existing CUDA codes with RISCV devices .…

## Selecting Optimal Trace Clustering Pipelines with AutoML

Trace clustering has been extensively used to preprocess event logs . Bygrouping similar behavior, these techniques guide the identification ofsub-logs, producing more understandable models and conformance analytics . Little attention has been posed to the relationship between eventlog properties and clustering quality .…

## Reference Publication Year Spectroscopy RPYS in practice Three RPYS analyzes in the course of Workshop III Cited References Analysis Using CRExplorer at the 18th ISSI conference

In course of the organization of Workshop III entitled “Cited ReferencesAnalysis Using CRExplorer” at the International Conference of the InternationalSociety for Scientometrics and Informetrics (ISSI2021) we have prepared threereference publication year spectroscopy (RPYS) analyzes . The three RPYS analyzes have shown quite differentseminal papers with a few overlaps .…

## Challenges in Generalization in Open Domain Question Answering

Recent work on Open Domain Question Answering has shown that there is a largediscrepancy in model performance between novel test questions and those that largely overlap with training questions . The strongest model performance forcomp-gen/novel-entity is 13.1/5.4% and 9.6/1.5% lower compared to that for the full test set .…

## Imposing Relation Structure in Language Model Embeddings Using Contrastive Learning

Language model text embeddings have revolutionized NLP research, but theirability to capture high-level semantic information is limited . In this paper, we propose a novel contrastivelearning framework that trains sentence . to encode the relations in agraph structure . Given a sentence (unstructured text) and its graph, we usecontrastive learning to impose relation-related structure .…

## ConQX Semantic Expansion of Spoken Queries for Intent Detection based on Conditioned Text Generation

ConQX is a method for semantic expansion of spoken queries . It uses the text generation ability of an auto-regressive language model, GPT-2 . To avoid off-topic text generation, we condition the input query to a structured context . We then apply zero-shot, one-shot and few-shot learning to fine-tune BERT and RoBERTa for intent detection .…

## A Novel Multi Centroid Template Matching Algorithm and Its Application to Cough Detection

Cough is a major symptom of respiratory-related diseases . Coughingcauses motion across the whole body and especially on the neck and head . Head motion data during coughing captured by a head-worn IMU sensor could be leveraged to detect coughs using a template matching algorithm .…

## DAG Oriented Protocols PHANTOM and GHOSTDAG under Incentive Attack via Transaction Selection Strategy

In response to the bottleneck of processing throughput inherent to singlechain PoW blockchains, several proposals have substituted a single chain for Directed Acyclic Graphs (DAGs) In this work, we investigate two notable DAG-oriented designs . We focus on PHANTOM (and its optimization GHOSTDAG), which proposes a custom transaction selection strategy that enables to increasethe throughput of the network .…

## Linked visualisations via Galois dependencies

We present new language-based dynamic analysis techniques for linking visualisations and other structured outputs to data in a fine-grained way . Our approach builds on bidirectionalprogram slicing techiques based on Galois connections, which provide desirableround-tripping properties . Unlike the prior work in program slicing, our approach allows selections tobe negated .…

## MACRPO Multi Agent Cooperative Recurrent Policy Optimization

This work considers the problem of learning cooperative policies in multi-agent settings with partially observable and non-stationary environments without a communication channel . We propose two novel ways of integrating information across agents and time in MACRPO . The code is available online athttps://://://github.com/kargarisaac/macrpo.…

## Self supervised Representation Learning for Trip Recommendation

Trip recommendation is a significant and engaging location-based service that can help new tourists make more customized travel plans . Conventional methods either leverage the heuristical algorithms (e.g., dynamic programming) or statistical analysis to search or rank a POI sequence .…

## Analysis of pseudo spectral methods used for numerical simulation of turbulence

Global spectral analysis (GSA) is used as a tool to test the accuracy of numerical methods with the help of canonical problems of convection and convection-diffusion equation . Forturbulent flows, an extreme event is characterized by the presence of lengthscales smaller than the Kolmogorov length scale, a heuristic limit for the largest wavenumber present without being converted to heat .…

## Complex Event Forecasting with Prediction Suffix Trees Extended Technical Report

Complex Event Recognition (CER) systems have become popular in the past twodecades due to their ability to “instantly” detect patterns on real-timestreams of events . However, there is a lack of methods for forecasting when apattern might occur before such an occurrence is actually detected by a CERengine .…

## A Weight Initialization Based on the Linear Product Structure for Neural Networks

Weight initialization plays an important role in training neural networks and affects tremendous deep learning applications . Various weightinitialization strategies have already been developed for different activation functions with different neural networks . These algorithms are based on minimizing the variance of the parameters between layers and might still fail when neural networks are deep, e.g.,…

## Prior Distribution Design for Music Bleeding Sound Reduction Based on Nonnegative Matrix Factorization

When we place microphones close to a sound source near other sources in audiorecording, the obtained audio signal includes undesired sound from the others . This is often called cross-talk or bleeding sound . For many audio applications including onstage sound reinforcement and sound editing after alive performance, it is important to reduce the bleeding sound in each recordedsignal .…

## Multipatch ZIKV Model and Simulations

Two multi-patch models for the spread of Zikavirus based on an SIRUV model . When the commuting between patches is ceased we expect that all the patches follow the dynamics of the single patch model . Weshow in an example that the effective population size should be used rather than the population size of the respective patch .…

## CTAL Pre training Cross modal Transformer for Audio and Language Representations

Existing audio-language task-specific predictive approaches focus on building complicated late-fusion mechanisms . We present a Cross-modal Transformer for Audio-and-Language, i.e.,CTAL, which aims to learn the intra-modality and inter-modalities connections between audio and language . We observe significant improvements across various tasks, such as, emotion classification, sentiment analysis, and speakerverification .…

## Variational Quantum Reinforcement Learning via Evolutionary Optimization

Recent advance in classical reinforcement learning (RL) and quantumcomputation points to a promising direction of performing RL on a quantumcomputer . Potential applications in quantum RL are limited by thenumber of qubits available in modern quantum devices . We present twoframeworks of deep quantum RL tasks using a gradient-free evolutionoptimization: First, we apply the amplitude encoding scheme to the Cart-Poleproblem .…