The author advocates two specific mathematical notations from his popularcourse and joint textbook, "Concrete Mathematics". The first of these,extending an idea of Iverson, is the notation "[P]"... | Continue reading
Data augmentation is a critical component of training deep learning models.Although data augmentation has been shown to significantly improve imageclassification, its potential has not been... | Continue reading
The Seventh International Workshop on Trends in Functional Programming inEducation, TFPIE 2018, was held on 14th of June 2018 at Chalmers University inGothenburg, Sweden, and was co-located with... | Continue reading
This paper reveals a trap for artificial general intelligence (AGI) theoristswho use economists' standard method of discounting. This trap is implicitly andfalsely assuming that a rational AGI... | Continue reading
When deploying autonomous agents in the real world, we need to think abouteffective ways of communicating our objectives to them. Traditional skilllearning has revolved around reinforcement and... | Continue reading
29 August 2018: "Artificial intelligence nails predictions of earthquakeaftershocks". This Nature News headline is based on the results of DeVries etal. (2018) who forecasted the spatial... | Continue reading
Medical imaging is crucial in modern clinics to guide the diagnosis andtreatment of diseases. Medical image reconstruction is one of the mostfundamental and important components of medical... | Continue reading
Developing agents that can quickly adapt their behavior to new tasks remainsa challenge. Meta-learning has been applied to this problem, but previousmethods require either specifying a reward... | Continue reading
In a commonly-used version of the Simple Assembly Line Balancing Problem(SALBP-1) tasks are assigned to stations along an assembly line with a fixedcycle time in order to minimize the required... | Continue reading
Modern generative models are usually designed to match target distributionsdirectly in the data space, where the intrinsic dimensionality of data can bemuch lower than the ambient... | Continue reading
This monograph attempts a theory of every 'thing' that can be distinguishedfrom other things in a statistical sense. The ensuing statisticalindependencies, mediated by Markov blankets, speak to... | Continue reading
This paper is a broad and accessible survey of the methods we have at ourdisposal for Monte Carlo gradient estimation in machine learning and across thestatistical sciences: the problem of... | Continue reading
This paper describes an advanced SQL injection technique where DNS resolutionprocess is exploited for retrieval of malicious SQL query results. ResultingDNS requests are intercepted by attackers... | Continue reading
There are anthropic reasons to suspect that life in more than three spatialdimensions is not possible, and if the same could be said of fewer than three,then one would have an anthropic argument... | Continue reading
Mixed Integer Programming (MIP) is one of the most widely used modelingtechniques to deal with combinatorial optimization problems. In manyapplications, a similar MIP model is solved on a... | Continue reading
Building multi-domain AI agents is a challenging task and an open problem inthe area of AI. Within the domain of dialog, the ability to orchestratemultiple independently trained dialog agents,... | Continue reading
Model-based reinforcement learning (RL) has proven to be a data efficientapproach for learning control tasks but is difficult to utilize in domains withcomplex observations such as images. In... | Continue reading
GraphQL is a novel query language proposed by Facebook to implement Web-basedAPIs. In this paper, we present a practical study on migrating API clients tothis new technology. First, we conduct a... | Continue reading
Virtually every connection to an Internet service is preceded by a DNSlookup. These lookups are performed in the clear without integrity protection,enabling manipulation, redirection,... | Continue reading
The ability to understand logical relationships between sentences is animportant task in language understanding. To aid in progress for this task,researchers have collected datasets for machine... | Continue reading
A continual learning agent learns online with a non-stationary andnever-ending stream of data. The key to such learning process is to overcomethe catastrophic forgetting of previously seen data,... | Continue reading
TV series such as HBO's most popular show Game of Thrones have seen a highnumber of dedicated followers, who watch and thoroughly analyze every minute ofthe show. Largely discussed aspect of the... | Continue reading
Sequential decision problems in applications such as manipulation inwarehouses, multi-step meal preparation, and routing in autonomous vehiclenetworks often involve reasoning about uncertainty,... | Continue reading
Wireless trajectory data consists of a number of (time, point) entries whereeach point is associated with a particular wireless device (WAP or BLE beacon)tied to a location identifier, such as a... | Continue reading
Advances in learning and representations have reinvigorated work thatconnects language to other modalities. A particularly exciting direction isVision-and-Language Navigation(VLN), in which... | Continue reading
In this work, we consider the problem of model selection for deepreinforcement learning (RL) in real-world environments. Typically, theperformance of deep RL algorithms is evaluated via... | Continue reading
Search engines are some of the most popular and profitable intelligenttechnologies in existence. Recent research, however, has suggested that searchengines may be surprisingly dependent on... | Continue reading
CNN-based methods have been proven to work well for saliency detection on RGBimages owing to the outstanding feature representation abilities of CNNs.However, their performance will degrade when... | Continue reading
Systematic comparison of methods for relation extraction (RE) is difficultbecause many experiments in the field are not described precisely enough to becompletely reproducible and many papers... | Continue reading
Generic Ontology Design Patterns, GODPs, are defined in Generic DOL, anextension of DOL, the Distributed Ontology, Model and Specification Language,and implemented using Heterogeneous Tool Set.... | Continue reading
The online sports gambling industry employs teams of data analysts to buildforecast models that turn the odds at sports games in their favour. Whileseveral betting strategies have been proposed... | Continue reading
Proposals for safe AGI systems are typically made at the level of frameworks,specifying how the components of the proposed system should be trained andinteract with each other. In this paper, we... | Continue reading
"Theorem proving is similar to the game of Go. So, we can probably improveour provers using deep learning, like DeepMind built the super-human computerGo program, AlphaGo." Such optimism has... | Continue reading
We present a novel extension to embedding-based knowledge graph completionmodels which enables them to perform open-world link prediction, i.e. topredict facts for entities unseen in training... | Continue reading
Explainability in Artificial Intelligence has been revived as a topic ofactive research by the need of conveying safety and trust to users in the `how'and `why' of automated decision-making.... | Continue reading
The MAPF problem is the fundamental problem of planning paths for multipleagents, where the key constraint is that the agents will be able to followthese paths concurrently without colliding... | Continue reading
Estimates of predictive uncertainty are important for accurate model-basedplanning and reinforcement learning. However, predictiveuncertainties---especially ones derived from modern deep... | Continue reading
Modern branch predictors predict the vast majority of conditional branchinstructions with near-perfect accuracy, allowing superscalar, out-of-orderprocessors to maximize speculative efficiency... | Continue reading
Existing performance comparisons of QUIC and TCP compared an optimized QUICto an unoptimized TCP stack. By neglecting available TCP improvementsinherently included in QUIC, comparisons do not... | Continue reading
Mutual information maximization has emerged as a powerful learning objectivefor unsupervised representation learning obtaining state-of-the-art performancein applications such as object... | Continue reading
State representation learning, or the ability to capture latent generativefactors of an environment, is crucial for building intelligent agents that canperform a wide variety of tasks. Learning... | Continue reading
This work revisits the use of information criteria to characterize thegeneralization of deep learning models. In particular, we empiricallydemonstrate the effectiveness of the Takeuchi... | Continue reading
Background: Evolution of open source projects frequently depends on a smallnumber of core developers. The loss of such core developers might bedetrimental for projects and even threaten their... | Continue reading
Unit testing has been considered as having a key role in building highquality software, and therefore it has been widely used in practice. However,data on the relationship between unit testing... | Continue reading
With the capability of modeling bidirectional contexts, denoisingautoencoding based pretraining like BERT achieves better performance thanpretraining approaches based on autoregressive language... | Continue reading
The reproduction and replication of novel results has become a major issuefor a number of scientific disciplines. In computer science and relatedcomputational disciplines such as systems... | Continue reading
As the performance of computer systems stagnates due to the end of Moore'sLaw, there is a need for new models that can understand and optimize theexecution of general purpose code. While there... | Continue reading