Transfer Learning for Human & AI

Continual learning has become increasingly important as it enables NLP modelsto constantly learn and gain knowledge over time . Our proposed method first disentangles text hidden spaces into representations that are generic to all tasks and representations specific to each individual task .…

Text style transfer aims to controllably generate text with targetedstylistic changes while maintaining core meaning from the source sentenceconstant . Many existing style transfer benchmarks do not offer fine-grained control on sentence structure, emphasis, and content of the sentence . We introduce a large-scale benchmark, StylePTB, with pairedsentences undergoing 21 stylistic changes spanning atomic lexical,syntactic, semantic, and thematic transfers of text .…

The UNESCO World Heritage List (WHL) is to identify the exceptionallyvaluable cultural and natural heritage to be preserved for mankind as a whole . The study applies state-of-the-art NLP models to build a classifier on a new real-world dataset containing officialOUV justification statements .…

A novel and efficient method is able to predict an additional 6.4% of facts in the LAMA benchmark . The training data used by these methodscontains certain regularities of the underlying fact distribution, and all the existing prompt methods, including ours, are able to exploit them for betterfact prediction .…

Programming errors in Ethereum smart contracts can result in catastrophic financial losses from stolen cryptocurrency . Indicators of Compromise (IoCs) are specially crafted for use in the blockchain . They need to capture the side-effects of successfulexploitation in the context of the Ethereum blockchain .…

Transformer attention architectures, similar to those developed for naturallanguage processing, have recently proved efficient also in vision . We report experiments with a simple such attentionsystem that can improve the performance of standard convolutional networks,with relatively few additional parameters . We demonstrate the usefulness of this network (GAttANet) for variousconvolutional backbones (from a simple 5-layer toy model to a standard ResNet50architecture) and datasets (CIFAR10, CIFAR100, Imagenet-1k) Each time, ourglobal attention system improves accuracy over the corresponding baseline .…

We propose RPSRNet – a novel end-to-end trainable deep neural network forrigid point set registration . For this task, we use a novel $2^D$-treerepresentation for the input point sets and a hierarchical deep featureembedding in the neural network . We achieve an inference speed of 12-15ms to register a pair of input pointclouds as large as 250K .…

Most users with at least 10 comments posted at least once on both left-leaning and right-leaning YouTube channels . We present the first large-scale measurement study of cross-partisandiscussions between liberals and conservatives on YouTube, based on a dataset of 274,241 political videos from 973 channels of US partisan media and 134Mcomments from 9.3M users over eight months in 2020 .…

Molecular communications is a promising framework for the design of controlled-release drug delivery systems . In a matrix carrier, drug molecules are dispersed inthe matrix and diffuse from the inner to the outer layers of the carrier onceimmersed in a dissolution medium .…

Terahertz (THz) communication is gaining more interest as one of the enablers of high-data-rate short-distance indoor applications inbeyond 5G networks . Non-orthogonal multiple-access (NOMA)-enabled schemes are promising schemes to realize the target spectral efficiency, lowlatency, and user fairness requirements in future networks .…

We present a SE(3)-equivariant graph neural network (GNN) approach that predicts the formation factor and effective permeability from micro-CT images . FFT solvers are established to compute both the formationfactor and effectiveness permeability, while the topology and geometry of the porespace are represented by a persistence-based Morse graph .…

Autopilot behavior can help to ensuresmoothhuman-vehicle collaboration during the initial exploration stagewith thevehicle . AutoPreview framework can provide a deeperunderstanding of autopilot behavior compared to direct interaction with the vehicle . We conducted acase study on humans-vehicles collaboration and built a prototype of our framework with theCARLA simulator .…

Z-numbers lack the ability to process the quantum information in quantum environment . The results show that, with the help of quantum computation, the proposed algorithm can make diagnoses correctly and efficiently . Based on QZNs, a novel quantummulti-attributes decision making (MADM) algorithm is proposed and applied in medical diagnosis.…

Attention-based encoder-decoder (AED) models learn an implicit internallanguage model (ILM) from the training transcriptions . Bayesian interpretation as in the hybrid autoregressivetransducer (HAT) suggests dividing by the prior of the discriminative acoustic model, which corresponds to this implicit LM . We propose several novel methods to estimate the ILM directly from the AED model .…

Robustness and counterfactual bias are usually evaluated on a test dataset . If the test dataset is perturbedslightly, will the evaluation results keep the same? In this paper, we propose a “double perturbation” framework to uncover model weaknesses beyond the test data .…

ENS has been criticized for its inherent design flaws, making the system vulnerable to kinds of attacks . DNS domain names are not fullycontrolled by the users, which can be easily taken down by the authorities andregistrars . Since blockchain has its unique properties like immutability anddecentralization, it seems to be promising to build a decentralized nameservice on blockchain .…

Deep learning methods have reached state-of-the-art performance in cardiac image segmentation . Currently, the main bottleneck towards their effective translation into clinics requires assuring continuous high model performance . In this work, we present a novel learning framework to monitor the performance of heart segmentation models in the absence of ground truth .…

Ourdescriptor, named MinkLoc++, can be used for place recognition, re-localization and loop closure purposes in robotics or autonomous vehicles applications . We introduce a discriminative multimodal descriptor based on a pair of sensorreadings: a point cloud from a LiDAR and an image from an RGB camera .…

Latest and current innovations of agricultural tech industry are increasingly driven by digital technologies . These digital farming solutions provide attractive advantages for farmers . The trend is going to devices and sensors, which send the acquired data directly to the cloud .…

We introduce the notion of universal graphs as a tool for constructing algorithms solving games of infinite duration such as parity games and meanpayoff games . We construct algorithms achieving or improving over the best known time and space complexity .…

Relay-assisted free-space optical (FSO) communication systems are exploited as a means to mitigate the limiting effects of the turbulence inducedatmospheric scintillation . Due to their mobility andflexibility, unmanned aerial vehicles (UAVs) provide new opportunities for FSOrelaying systems . In this paper, a hovering UAV-based serial FSOdecode-and-forward relaying system is investigated .…

GenerativeAdversarial Networks (GANs) involve training of overparameterized models where the number of parameters of the model exceeds a certain threshold . The role and benefits of modeloverparameterization in the convergence of GDA to a global saddle point in non-convex concave min-max problems is far less understood .…

A multitude of classifiers can be trained on the same data to achieve similar performances during test time . This phenomenon is often associated with the blind selection of one model instead of another with similar performances . A model-agnostic algorithm, DIG, is proposed to capture and explaindiscrepancies locally .…

Data-driven artificial intelligence models fed with published scientific findings have been used to create powerful prediction engines for scientific and technological advance . These models succeed by predicting human predictions and the scientists who will make them . By tuning AI to avoid the crowd, however, itgenerates scientifically promising “alien” hypotheses unlikely to be imagined or pursued without intervention, not only accelerating but punctuatingscientific advance .…

State-of-the-art deep neural networks require large-scale labeled training data that is often expensive to obtain or not available for many tasks . In this work, we develop a weak supervision framework (ASTRA) that leveragesall the available data for a given task .…

SuperSim is alarge-scale similarity and relatedness test set for Swedish built with experthuman judgments . The test set is composed of 1,360 word-pairs independentlyjudged for both relatedness and similarity by five annotators . We evaluate different models (Word2Vec, fastText, and GloVe) trained on two separate datasets (Swedish Gigaword and Swedish Wikipediadump) to provide a baseline for future comparison .…

The Cyber Science Lab (CSL) and Smart Cyber-Physical System (SCPS) Lab at theUniversity of Guelph conduct a market study of cybersecurity technologyadoption and requirements for smart and precision farming in Canada . Weconducted 17 stakeholder/key opinion leader interviews in Canada and the USA, to complete this study .…

2D image-based virtual try-on has attracted increased attention from themultimedia and computer vision communities . We propose a novel two-stage Cloth Interactive Transformer (CIT) for virtualtry-on . In the first stage, we design a CIT matching block, aiming to perform alearnable thin-plate spline transformation that can capture more reasonable long-range relation .…

X-ray imaging is a widely used technique for non-destructive inspection of agricultural food products . A foreign object is defined as a fragment of material with differentX-ray attenuation properties than those belonging to the food product . A novelthickness correction model is introduced as a pre-processing technique for DEXAdata .…

Index plays an essential role in modern database engines to accelerate the processing . LIPP is able to support all kinds of index operations, namely lookup query,range query, insert, delete, update and bulkload . The results demonstrate that our method consistentlyoutperforms state-of-the-art solutions, achieving by up to 4x for a broaderclass of workloads with different index operations .…

Modern 3D printing technologies and the upcoming mass-customization paradigm call for efficient methods to produce and distribute arbitrarily-shaped 3D objects . This paper introduces an original algorithm to split a 3D model inparts that can be efficiently packed within a box .…

The problem of the real-time multiple-input multiple- input multiple-output (MIMO) arraycontrol is addressed by means of an innovative hybrid beamforming technique . The synthesis of the excitationsof the MIMO system is first re-formulated as that of matching an ideal “hybrid” pattern fitting capacity or low-interference constraints along the angularcoordinates .…

In mobile games industry, the churn rate is often pronounced due to the high competition and cost in customer acquisition . This needs churn prediction, predicting users who will be churning within a given time period . Accuratechurn prediction can enable the businesses to devise and engage strategicremediations to maintain a low churn rate.…

Trust in algorithms can be established by letting users explore results and find patterns . Reflecting features and patterns of human understanding of a domain against algorithmic results can increase the trust that a user has inthe solution . By either accepting or rejecting a solution, the user’sset of expectations evolves and a learning process for the users is established .…

Long-term network-wide accurate traffic speed forecasting is one of the most challenging tasks . Most proposed models treat it as a spatiotemporal graph modeling problem and use Graph Convolution Network (GCN) based methods . This paper proposes a novel model, Traffic Transformer, for spatial-temporal graph models and long-termtraffic forecasting .…

We describe our straight-forward approach for Tasks 5 and 6 of 2021 SocialMedia Mining for Health Applications (SMM4H) shared tasks . We explore how much fine-tuning is necessary for classifying tweets as containing self-reported COVID-19 symptoms (Task 5) or whether a tweet related to the virus is self-reporting, non-personalreporting, or a literature/news mention of the virus (Task 6)…

Human ratings are one of the most prevalent methods to evaluate the performance of natural language processing algorithms . It is commonto measure the quality of sentences generated by a natural language generationmodel using human raters . In this paper, we argue for exploring the use ofsubjective evaluations within the process of training language generationmodels in a multi-task learning setting .…

We study how to introduce locality mechanisms into vision transformers . Locally-enhanced transformers outperform the baselines DeiT-T and PVT-t by 2.6\% and 3.1\% with a negligible increase in the number of parameters and effort . The same mechanism was used to apply to 4 visiontransformers, which shows the generalization of the locality concept .…

The photo-realism of images synthesized by GenerativeAdversarial Networks (GANs) has gone from somewhat reasonable to almost perfect . This trajectory hasled many of the state-of-the-art GANs to be inaccessibly large, disengagingmany without large computational resources . We present an unsupervised method to find a direction in the latentspace that aligns with improved photo realism.…

Aregularization term aims to maximize correlation between the rankings of the shared-weight network and that of the standalonearchitectures . We incorporate our regularization term into three different NAS algorithms . We show that it consistently improves performance across algorithms, search-spaces, and tasks, and shows that it can be used to improve search performance on other parts of the network .…

Indoor venues accommodate many people who collectively form crowds . People may prefer to avoid crowded rooms when walking from A to B . Indoor Least Crowded Path Query (LCPQ) finds a path with the shortest travel time in the presence of crowds .…

We propose a computational design tool to enable casual end-users to easilydesign, fabricate, and assemble flat-pack furniture with guaranteedmanufacturability . Using our system, users select parameterized components from a library and constrain their dimensions . Then they abstractly specifyconnections among components to define the furniture .…

We consider a general statistical inference model of finite-rank tensorproducts . For any interaction structure and any order of tensor products, we identify the limit free energy of the model in terms of a variational formula . We show that the free energy must be the solution to a certain Hamilton-Jacobi equation .…

We take the first step in using vehicle-to-vehicle (V2V) communication toprovide real-time on-board traffic predictions . In order to best utilizereal-world V2V communication data, we integrate first principle models with deep learning techniques . Our approach is able to predict velocity of individual vehicles up to a minute into the future with improved accuracy over first principle-based baselines .…

The open-domain sketch-to-photo translation is challenging due to the lack of training supervision and the large geometrydistortion between the freehand sketch and photo domains . We propose a framework that jointlylearns sketch to photo and photo-to sketch generation . We validate our method on the Scribble and SketchyCOCO datasets.…

The QLattice is a supervised machine learning tool inspired by Richard Feynman’spath integral formulation, that explores many potential models that solves agiven problem . It formulates these models as graphs that can be interpreted as equations . We show how it differs from traditional machine learning approaches, what it has in common with them .…

Deep learning recommendation models (DLRMs) are used across many business-critical services at Facebook and are the single largest AIapplication in terms of infrastructure demand in its data-centers . We introduce a high-performancescalable software stack based on PyTorch and pair it with the new evolution of the\zion platform, namely \zionex .…

Expressive reading, considered the defining attribute of oral readingfluency, comprises the prosodic realization of phrasing and prominence . We consider a labeled dataset of children’s readingrecordings for the speaker-independent detection of prominent words using acoustic-prosodic and lexico-syntactic features . Deeplearning is applied to obtain word-level features from low-level acousticcontours of fundamental frequency, intensity and spectral shape in anend-to-end fashion .…

The current state-of-the-art model HiAGM for hierarchical text classification has two limitations . We propose HTCInfoMax to addressthese issues by introducing information maximization which includes twomodules . The first module can model interaction between each text sample and its ground truth labels explicitly which filters out irrelevant information .…

In this paper, we aim at improving the computational efficiency of graphconvolutional networks (GCNs) for learning on point clouds . The optimized networks have reduced computational complexity, decreased memoryconsumption, and accelerated inference speed . Code will be available at\url{https://://://github.com/ofsoundof/EfficientGCN.git and the code is available at http://://www.gofsoundsof-of-soundof.com.org//gofEfficient…