Finding the seed set that maximizes the influence spread over a network is a well-known NP-hard problem. Though a greedy algorithm can provide near-optimal solutions, the subproblem of influence estimation renders the solutions inefficient. In this work, we propose GLIE, a graph neural network that learns how to estimate the influence spread of the independent cascade. GLIE relies on a theoretical upper bound that is tightened through supervised training. Experiments indicate that it provides accurate influence estimation for real graphs up to 10 times larger than the train set. Subsequently, we incorporate it into two influence maximization techniques. We first utilize Cost Effective Lazy Forward optimization substituting Monte Carlo simulations with GLIE, surpassing the benchmarks albeit with a computational overhead. To improve computational efficiency we develop a provably submodular influence spread based on GLIE's representations, to rank nodes while building the seed set adaptively. The proposed algorithms are inductive, meaning they are trained on graphs with less than 300 nodes and up to 5 seeds, and tested on graphs with millions of nodes and up to 200 seeds. The final method exhibits the most promising combination of time efficiency and influence quality, outperforming several baselines.

A huge number of display advertising campaigns are launched every day by advertisers in order to promote their products or services. The main objective of each advertiser is to display ads to specific groups of users, i.e. users who meet specific criteria or their interests are related to the promoted products or services. Audience expansion, also known as audience look-alike targeting, is one of the major display advertising techniques that helps advertisers to discover audiences with similar attributes to a target audience who is interested in advertisers' products or services. In this paper, we present different audience expansion schemes able to identify users with similar browsing interests to those of the seed users provided by the advertiser. The proposed audience expansion schemes are based on different unsupervised representation models that are able to capture the interests of the users according to their browsing history. We have conducted an extensive empirical study on a real data collected from an advertising platform to analyse the effectiveness of the proposed schemes to expand the audiences of five different advertisers.

International audience

Information diffusion and social influence are more and more present in today's Web ecosystem. Having algorithms that optimize the presence and message diffusion on social media is indeed crucial to all actors (media companies, political parties, corporations, etc.) who advertise on the Web. Motivated by the need for effective viral marketing strategies, influence estimation and influence maximization have therefore become important research problems, leading to a plethora of methods. However, the majority of these methods are non-adaptive, and therefore not appropriate for scenarios in which influence campaigns may be ran and observed over multiple rounds, nor for scenarios which cannot assume full knowledge over the diffusion networks and the ways information spreads in them.

This paper examines the problem of adaptive influence maximization in social networks. As adaptive decision making is a time-critical task, a realistic feedback model has been considered, called myopic. In this direction, we propose the myopic adaptive greedy policy that is guaranteed to provide a (1 - 1/e) -approximation of the optimal policy under a variant of the independent cascade diffusion model. This strategy maximizes an alternative utility function that has been proven to be adaptive monotone and adaptive submodular. The proposed utility function considers the cumulative number of active nodes through the time, instead of the total number of the active nodes at the end of the diffusion. Our empirical analysis on real-world social networks reveals the benefits of the proposed myopic strategy, validating our theoretical results.

In this paper, we introduce AngryBER, an intelligent agent architecture on the Angry Birds domain that employs a Bayesian ensemble inference mechanism to promote decision-making abilities. It is based on an efficient tree-like structure for encoding and representing game screenshots, where it exploits its enhanced modeling capabilities. This has the advantage to establish an informative feature space and translate the task of game playing into a regression analysis problem. A Bayesian ensemble regression framework is presented by considering that every combination of objects' material and bird type has its own regression model. We address the problem of action selection as a multiarmed bandit problem, where the upper confidence bound (UCB) strategy has been used. An efficient online learning procedure has been also developed for training the regression models. We have evaluated the proposed methodology on several game levels, and compared its performance with published results of all agents that participated in the 2013 and 2014 Angry Birds AI competitions. The superiority of the new method is readily deduced by inspecting the reported results.

We introduce a simple, general framework for likelihood-free Bayesian reinforcement learning, through Approximate Bayesian Computation (ABC). The advantage is that we only require a prior distribution on a class of simulators. This is useful when a probabilistic model of the underlying process is too complex to formulate, but where detailed simulation models are available. ABC-RL allows the use of any Bayesian reinforcement learning technique in this case. It can be seen as an extension of simulation methods to both planning and inference. We experimentally demonstrate the potential of this approach in a comparison with LSPI. Finally, we introduce a theorem showing that ABC is sound.

A significant issue in representing reinforcement learning agents in Markov decision processes is how to design efficient feature spaces in order to estimate optimal policy. This particular study addresses this challenge by proposing a compact framework that employs an on-line clustering approach for constructing appropriate basis functions. Also, it performs a state-action trajectory analysis to gain valuable affinity information among clusters and estimate their transition dynamics. Value function approximation is used for policy evaluation in a least-squares temporal difference framework. The proposed method is evaluated in several simulated and real environments, where we took promising results.

In this work we present an advanced Bayesian formulation to the task of control learning that employs the Relevance Vector Machines (RVM) generative model for value function evaluation. The key aspect of the proposed method is the design of the discount return as a generalized linear model that constitutes a well-known probabilistic approach. This allows to augment the model with advantageous sparse priors provided by the RVM's regression framework. We have also taken into account the significant issue of selecting the proper parameters of the kernel design matrix. Experiments have shown that our method produces improved performance in both simulated and real test environments.