Serverless computing (also known as functions as a service) is a new cloudcomputing abstraction that makes it easier to write robust, large-scale webservices. In serverless computing,... | Continue reading
This paper explores the task Natural Language Understanding (NLU) by lookingat duplicate question detection in the Quora dataset. We conducted extensiveexploration of the dataset and used... | Continue reading
We humans seem to have an innate understanding of the asymmetric progressionof time, which we use to efficiently and safely perceive and manipulate ourenvironment. Drawing inspiration from that,... | Continue reading
C programs can use compiler builtins to provide functionality that the Clanguage lacks. On Linux, GCC provides several thousands of builtins that arealso supported by other mature compilers,... | Continue reading
The Digital Ludeme Project (DLP) aims to reconstruct and analyse over 1000traditional strategy games using modern techniques. One of the key aspects ofthis project is the development of Ludii, a... | Continue reading
Continual learning (CL) is a particular machine learning paradigm where thedata distribution and learning objective changes through time, or where all thetraining data and objective criteria are... | Continue reading
Extracting fashion attributes from images of people wearing clothing/fashionaccessories is a very hard multi-class classification problem. Most often, evencatalogues of fashion do not have all... | Continue reading
Multi-hop Reading Comprehension (RC) requires reasoning and aggregationacross several paragraphs. We propose a system for multi-hop RC that decomposesa compositional question into simpler... | Continue reading
Our work presents extensive empirical evidence that layer rotation, i.e. theevolution across training of the cosine distance between each layer's weightvector and its initialization, constitutes... | Continue reading
Our work presents extensive empirical evidence that layer rotation, i.e. theevolution across training of the cosine distance between each layer's weightvector and its initialization, constitutes... | Continue reading
Sequence prediction models can be learned from example sequences with avariety of training algorithms. Maximum likelihood learning is simple andefficient, yet can suffer from compounding error... | Continue reading
We deal with the \textit{selective classification} problem(supervised-learning problem with a rejection option), where we want to achievethe best performance at a certain level of coverage of... | Continue reading
Different linguistic perspectives cause many diverse segmentation criteriafor Chinese word segmentation (CWS). Most existing methods focus on improvingthe performance of single-criterion CWS.... | Continue reading
In this work we propose a non-dimensionalization approach for theStokes-Brinkman model for flow in porous media. We study the effect of thedimensionless number found, which will be denoted by A... | Continue reading
The most famous single-player card game is 'Klondike', but our ignorance ofits winnability percentage has been called "one of the embarrassments ofapplied mathematics". Klondike is just one of... | Continue reading
Advanced driver assistance systems (ADASs) were developed to reduce thenumber of car accidents by issuing driver alert or controlling the vehicle. Inthis paper, we tested the robustness of... | Continue reading
Cities are systems with a large number of constituents and agents interactingwith each other and can be considered as emblematic of complex systems.Modeling these systems is a real challenge and... | Continue reading
Anticipatory thinking is a complex cognitive process for assessing andmanaging risk in many contexts. Humans use anticipatory thinking to identifypotential future issues and proactively take... | Continue reading
Data selection methods such as active learning and core-set selection areuseful tools for machine learning on large datasets, but they can beprohibitively expensive to apply in deep learning.... | Continue reading
We share the findings of the first shared task on improving robustness ofMachine Translation (MT). The task provides a testbed representing challengesfacing MT models deployed in the real world,... | Continue reading
Neural architectures inspired by our own human cognitive system, such as therecently introduced world models, have been shown to outperform traditionaldeep reinforcement learning (RL) methods in... | Continue reading
Blockchain is a promising technology for establishing trust in IoT networks,where network nodes do not necessarily trust each other. Cryptographic hashlinks and distributed consensus mechanisms... | Continue reading
Clustering and analyzing on collected data can improve user experiences andquality of services in big data, IoT applications. However, directly releasingoriginal data brings potential privacy... | Continue reading
In this article, we provide a summary of recent efforts towards achievingInternet geolocation securely, \ie without allowing the entity being geolocatedto cheat about its own geographic... | Continue reading
IP headers include a 16-bit ID field. Our work examines the generation ofthis field in Windows (versions 8 and higher), Linux and Android, and showsthat the IP ID field enables remote servers to... | Continue reading
DNS is a distributed, fault tolerant system that avoids a single point offailure. As such it is an integral part of the internet as we use it today andhence deemed a safe protocol which is let... | Continue reading
We investigate the capacity of mechanisms for compositional semantic parsingto describe relations between sentences and semantic representations. We prove that in order to represent certain... | Continue reading
Most semantic parsers that map sentences to graph-based meaningrepresentations are hand-designed for specific graphbanks. We present acompositional neural semantic parser which achieves, for the... | Continue reading
In the last five years, private companies, research institutions as well aspublic sector organisations have issued principles and guidelines for ethicalAI, yet there is debate about both what... | Continue reading
This technical note describes a set of baseline tools for automaticprocessing of Danish text. The tools are machine-learning based, using naturallanguage processing models trained over... | Continue reading
We propose a contextual emotion classifier based on a transferable languagemodel and dynamic max pooling, which predicts the emotion of each utterance ina dialogue. A representative emotion... | Continue reading
We use the English model of BERT and explore how a deletion of one word in asentence changes representations of other words. Our hypothesis is thatremoving a reducible word (e.g. an adjective)... | Continue reading
We present a study of morphological irregularity. Following recent work, wedefine an information-theoretic measure of irregularity based on thepredictability of forms in a language. Using a... | Continue reading
Chinese word segmentation (CWS) is a fundamental step of Chinese naturallanguage processing. In this paper, we build a new toolkit, named PKUSEG, formulti-domain word segmentation. Unlike... | Continue reading
Public debate forums provide a common platform for exchanging opinions on atopic of interest. While recent studies in natural language processing (NLP)have provided empirical evidence that the... | Continue reading
Knowledge graphs store facts using relations between pairs of entities. Inthis work, we address the question of link prediction in knowledge bases whereeach relation is defined on any number of... | Continue reading
Treebanks traditionally treat punctuation marks as ordinary words, butlinguists have suggested that a tree's "true" punctuation marks are notobserved (Nunberg, 1990). These latent "underlying"... | Continue reading
Scientific models describe natural phenomena at different levels ofabstraction. Abstract descriptions can provide the basis for interventions onthe system and explanation of observed phenomena... | Continue reading
Dempster-Shafer evidence theory has been widely used in various fields ofapplications, because of the flexibility and effectiveness in modelinguncertainties without prior information. However,... | Continue reading
To solve tasks in new environments involving objects unseen during training,agents must reason over prior information about those objects and theirrelations. We introduce the Prior Knowledge... | Continue reading
Fuzzy cognitive maps (FCMs) model feedback causal relations in interwovenwebs of causality and policy variables. FCMs are fuzzy signed directed graphsthat allow degrees of causal influence and... | Continue reading
Score matching is a popular method for estimating unnormalized statisticalmodels. However, it has been so far limited to simple, shallow models orlow-dimensional data, due to the difficulty of... | Continue reading
There is substantial interest in the use of machine learning (ML) basedtechniques throughout the electronic computer-aided design (CAD) flow,particularly those based on deep learning. However,... | Continue reading
Humans do not make inferences over texts, but over models of what texts areabout. When annotators are asked to annotate coreferent spans of text, it istherefore a somewhat unnatural task. This... | Continue reading
Ontology-based knowledge bases (KBs) like DBpedia are very valuableresources, but their usefulness and usability is limited by various qualityissues. One such issue is the use of string literals... | Continue reading
Multiagent decision-making problems in partially observable environments areusually modeled as either extensive-form games (EFGs) within the game theorycommunity or partially observable... | Continue reading
This paper aims to question the suitability of the Turing Test, for testingmachine intelligence, in the light of advances made in the last 60 years inscience, medicine, and philosophy of mind.... | Continue reading
Decision making in dense traffic can be challenging for autonomous vehicles.An autonomous system only relying on predefined road priorities and consideringother drivers as moving objects will... | Continue reading