Search

Large Language Models work quite well with general-purpose data and many tasks in Natural Language Processing. However, they show several limitations when used for a task such as domain-specific abstractive text summarization. This paper identifies three of those limitations as research problems in the context of abstractive text summarization: 1) Quadratic complexity of transformer-based models with respect to the input text length; 2) Model Hallucination, which is a model's ability to generate factually incorrect text; and 3) Domain Shift, which happens when the distribution of the model's training and test corpus is not the same. Along with a discussion of the open research questions, this paper also provides an assessment of existing state-of-the-art techniques relevant to domain-specific text summarization to address the research gaps.

In this paper we investigate the board game Cram, which isan impartial combinatorial game, using an alpha-beta solver. Since Cram is anotoriously hard game in the sense that it is difficult to obtain reliableand useful domain knowledge to use in the search process, we decided torely on knowledge obtained from Combinatorial Game Theory (CGT).The first and most effective addition to our solver is to incorporateendgame databases prefilled with CGT values (nimbers) for all positionsfitting on boards with at most 30 squares. This together with twoefficient move-ordering heuristics aiming at early splitting positions intofragments fitting in the available databases gives a large improvement ofsolving power. Next we define five more heuristics based on CGT thatcan be used to further reduce the sizes of the search trees considerably.In the final version of our program we were able to solve all odd x oddCram boards for which results were available from the literature (evenx even and odd x even boards are trivially solved). Investigating newboards led to solving two boards not solved before, namely the 3 x 21board, a first-player win, and the 5 x 11 board, a second-player win.

Detecting false information in the form of fake news has become a bigger challenge than anticipated. There are multiple promising ways of approaching such a problem, ranging from source-based detection, linguistic feature extraction, and sentiment analysis of articles. While analyzing the sentiment of text has produced some promising results, this paper explores a rather more fine-grained strategy of classifying news as fake or real, based solely on the emotion profile of an article's title. A RoBERTa model was first trained to perform Emotion Classification, achieving test accuracy of about 90%. Six basic emotions were used for the task, based on the prominent psychologist Paul Ekman - fear, joy, anger, sadness, disgust and surprise. A seventh emotional category was also added to represent neutral text. Model performance was also validated by comparing classification results to other state-of-the-art models, developed by other groups. The model was then used to make inference on the emotion profile of news titles, returning a probability vector, which describes the emotion that the title conveys. Having the emotion probability vectors for each article's title, another Binary Random Forest classifier model was trained to evaluate news as either fake or real, based solely on their emotion profile. The model achieved up to 88% accuracy on the Kaggle Fake and Real News Dataset, showing there is a connection present between the emotion profile of news titles and if the article is fake or real.

Agent-oriented solutions form a useful paradigm to design intelligent systems. For health-related behaviour change, this is also a promising approach. Designing an agent for lifestyle change interventions is a difficult task because socio-ecological models are involved that represent many conflicting desires and goals. Different types of cognitive architectures are available to design this type of health behavior agents but they are rarely used. In this paper, we used the BDI model to design a health behavior agent that will execute behavior change intervention for a better healthy lifestyle. We explain the working of the architecture by the example of an agent which uses adaptive goals-setting and a percentile scheduling technique for increasing physical activity.

In Reinforcement learning, Q-learning is the best-known algorithm but it suffers from overestimation bias, which may lead to poor performance or unstable learning. In this paper, we present a novel analysis of this problem using various control tasks. For solving these tasks, Q-learning is combined with a multilayer perceptron (MLP), experience replay, and a target network. We focus our analysis on the effect of the learning rate when training the MLP. Furthermore, we examine if decaying the learning rate over time has advantages over static ones. Experiments have been performed using various maze-solving problems involving deterministic or stochastic transition functions and 2D or 3D grids and two Open-AI gym control problems. We conducted the same experiments with Double Q-learning using two MLPs with the same parameter settings, but without target networks. The results on the maze problems show that for Q-learning combined with the MLP, the overestimation occurs when higher learning rates are used and not when lower learning rates are used. The Double Q-learning variant becomes much less stable with higher learning rates and with low learning rates the overestimation bias may still occur. Overall, decaying learning rates clearly improves the performances of both Q-learning and Double Q-learning.

Q-learning is a reinforcement learning algorithm that has overestimation bias, because it learns the optimal action values by using a target that maximizes over uncertain action-value estimates. Although the overestimation bias of Q-learning is generally considered harmful, a recent study suggests that it could be either harmful or helpful depending on the reinforcement learning problem. In this paper, we propose a new Q-learning variant, called Variation-resistant Q-learning, to control and utilize estimation bias for better performance. Firstly, we present the tabular version of the algorithm and mathematically prove its convergence. Secondly, we combine the algorithm with function approximation. Finally, we present empirical results from three different experiments, in which we compared the performance of Variation-resistant Q-learning, Q-learning, and Double Q-learning. The empirical results show that Variation-resistant Q-learning can control and utilize estimation bias for better performance in the experimental tasks.

Harmony in visual compositions is a concept that cannot be defined or easily expressed mathematically, even by humans. The goal of the research described in this paper was to find a numerical representation of artistic compositions with different levels of harmony. We ask humans to rate a collection of grayscale images based on the harmony they convey. To represent the images, a set of special features were designed and extracted. By doing so, it became possible to assign objective measures to subjectively judged compositions. Given the ratings and the extracted features, we utilized machine learning algorithms to evaluate the efficiency of such representations in a harmony classification problem. The best performing model (SVM) achieved 80% accuracy in distinguishing between harmonic and disharmonic images, which reinforces the assumption that concept of harmony can be expressed in a mathematical way that can be assessed by humans., Comment: 9 pages, ICAART 2021

The extraction of named entities from unstructured text is a crucial component in numerous Natural Language Processing (NLP) applications such as information retrieval, question answering, machine translation, to name but a few. Named-entity Recognition (NER) aims at locating proper nouns from unstructured text and classifying them into a predefined set of types, such as persons, locations, and organizations. There has been extensive research on improving the accuracy of NER in English text. For other languages such as Arabic, extracting Named-entities is quite challenging due to its morphological structure. In this paper, we introduce ArabiaNer, a system employing Conditional Random Field (CRF) learning algorithm with extensive feature engineering steps to effectively extract Arabic named Entities. ArabiaNer produced state-of-the-art results with f1-score of 91.31% when applied on the ANERcrop dataset.

Deep Reinforcement Learning (DRL) has the potential to surpass the existing state of the art in various practical applications. However, as long as learned strategies and performed decisions are difficult to interpret, DRL will not find its way into safety-relevant fields of application. SHAP values are an approach to overcome this problem. It is expected that the addition of these values to DRL provides an improved understanding of the learned action-selection policy. In this paper, the application of a SHAP method for DRL is demonstrated by means of the OpenAI Gym LongiControl Environment. In this problem, the agent drives an autonomous vehicle under consideration of speed limits in a single lane route. The controls learned with a DDPG algorithm are interpreted by a novel approach combining learned actions and SHAP values. The proposed RL-SHAP representation makes it possible to observe in every time step which features have a positive or negative effect on the selected action and which influences are negligible. The results show that RL-SHAP values are a suitable approach to interpret the decisions of the agent.

In the quest for efficient and robust learning methods, combining unsupervised state representation learning and reinforcement learning (RL) could offer advantages for scaling RL algorithms by providing the models with a useful inductive bias. For achieving this, an encoder is trained in an unsupervised manner with two state representation methods, a variational autoencoder and a contrastive estimator. The learned features are then fed to the actor-critic RL algorithm Proximal Policy Optimization (PPO) to learn a policy for playing Open AI's car racing environment. Hence, such procedure permits to decouple state representations from RL-controllers. For the integration of RL with unsupervised learning, we explore various designs for variational autoencoders and contrastive learning. The proposed method is compared to a deep network trained directly on pixel inputs with PPO. The results show that the proposed method performs slightly worse than directly learning from pixel inputs; however, it has a more stable learning curve, a substantial reduction of the buffer size, and requires optimizing 88% fewer parameters. These results indicate that the use of pre-trained state representations has several benefits for solving RL tasks.

This paper casts coordination of a team of robots within the framework of game theoretic learning algorithms. In particular a novel variant of fictitious play is proposed, by considering multi-model adaptive filters as a method to estimate other players’ strategies. The proposed algorithm can be used as a coordination mechanism between players when they should take decisions under uncertainty. Each player chooses an action after taking into account the actions of the other players and also the uncertainty. Uncertainty can occur either in terms of noisy observations or various types of other players. In addition, in contrast to other game-theoretic and heuristic algorithms for distributed optimisation, it is not necessary to find the optimal parameters a priori. Various parameter values can be used initially as inputs to different models. Therefore, the resulting decisions will be aggregate results of all the parameter values. Simulations are used to test the performance of the proposed methodology against other game-theoretic learning algorithms.

The practical availability of Audiovisual Processing tools to media scholars and heritage institutions remains limited, despite all the technical advancements of recent years. In this article we present the approach chosen in the CLARIAH project to increase this availability, we discuss the challenges encountered, and introduce the technical solutions we are implementing. Through three use cases focused on the enrichment of AV archives, Pose Analysis, and Automatic Speech Recognition, we demonstrate the potential and breadth of using Audiovisual Processing for archives and Digital Humanities research.

Demonstrating reliability of cognitive multi-agent systems is of key importance. There has been an extensive amount of work on logics for verifying cognitive agents but it has remained mostly theoretical. Cognitive agent-oriented programming languages provide the tools for compact representation of complex decision making mechanisms, which offers an opportunity for applying a theorem proving approach. We base our work on the belief that theorem proving can add to the currently available approaches for providing assurance for cognitive multi-agent systems. However, a practical approach using theorem proving is missing. We explore the use of proof assistants to make verifying cognitive multi-agent systems more practical.