Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

"Recommendation": models, code, and papers

Rotting Bandits

Nov 02, 2017
Nir Levine, Koby Crammer, Shie Mannor

The Multi-Armed Bandits (MAB) framework highlights the tension between acquiring new knowledge (Exploration) and leveraging available knowledge (Exploitation). In the classical MAB problem, a decision maker must choose an arm at each time step, upon which she receives a reward. The decision maker's objective is to maximize her cumulative expected reward over the time horizon. The MAB problem has been studied extensively, specifically under the assumption of the arms' rewards distributions being stationary, or quasi-stationary, over time. We consider a variant of the MAB framework, which we termed Rotting Bandits, where each arm's expected reward decays as a function of the number of times it has been pulled. We are motivated by many real-world scenarios such as online advertising, content recommendation, crowdsourcing, and more. We present algorithms, accompanied by simulations, and derive theoretical guarantees.


  Access Paper or Ask Questions

On the challenges of learning with inference networks on sparse, high-dimensional data

Oct 17, 2017
Rahul G. Krishnan, Dawen Liang, Matthew Hoffman

We study parameter estimation in Nonlinear Factor Analysis (NFA) where the generative model is parameterized by a deep neural network. Recent work has focused on learning such models using inference (or recognition) networks; we identify a crucial problem when modeling large, sparse, high-dimensional datasets -- underfitting. We study the extent of underfitting, highlighting that its severity increases with the sparsity of the data. We propose methods to tackle it via iterative optimization inspired by stochastic variational inference \citep{hoffman2013stochastic} and improvements in the sparse data representation used for inference. The proposed techniques drastically improve the ability of these powerful models to fit sparse data, achieving state-of-the-art results on a benchmark text-count dataset and excellent results on the task of top-N recommendation.

* 14 pages, 3 tables, 11 figures 

  Access Paper or Ask Questions

Movie Popularity Classification based on Inherent Movie Attributes using C4.5,PART and Correlation Coefficient

Sep 26, 2012
Khalid Ibnal Asad, Tanvir Ahmed, Md. Saiedur Rahman

Abundance of movie data across the internet makes it an obvious candidate for machine learning and knowledge discovery. But most researches are directed towards bi-polar classification of movie or generation of a movie recommendation system based on reviews given by viewers on various internet sites. Classification of movie popularity based solely on attributes of a movie i.e. actor, actress, director rating, language, country and budget etc. has been less highlighted due to large number of attributes that are associated with each movie and their differences in dimensions. In this paper, we propose classification scheme of pre-release movie popularity based on inherent attributes using C4.5 and PART classifier algorithm and define the relation between attributes of post release movies using correlation coefficient.

* IEEE/OSA/IAPR International Conference on Informatics, Electronics & Vision (ICIEV2012), pp. 747-752, May 2012 
* 6 pages 

  Access Paper or Ask Questions

A global physician-oriented medical information system

Oct 11, 2008
Axel Boldt, Michael Janich

We propose to improve medical decision making and reduce global health care costs by employing a free Internet-based medical information system with two main target groups: practicing physicians and medical researchers. After acquiring patients' consent, physicians enter medical histories, physiological data and symptoms or disorders into the system; an integrated expert system can then assist in diagnosis and statistical software provides a list of the most promising treatment options and medications, tailored to the patient. Physicians later enter information about the outcomes of the chosen treatments, data the system uses to optimize future treatment recommendations. Medical researchers can analyze the aggregate data to compare various drugs or treatments in defined patient populations on a large scale.

* 8 pages 

  Access Paper or Ask Questions

Surveying 5G Techno-Economic Research to Inform the Evaluation of 6G Wireless Technologies

Jan 10, 2022
Edward J. Oughton, William Lehr

Techno-economic assessment is a fundamental technique engineers use for evaluating new communications technologies. However, despite the techno-economics of the fifth cellular generation (5G) being an active research area, it is surprising there are few comprehensive evaluations of this growing literature. With mobile network operators deploying 5G across their networks, it is therefore an opportune time to appraise current accomplishments and review the state-of-the-art. Such insight can inform the flurry of 6G research papers currently underway and help engineers in their mission to provide affordable high-capacity, low-latency broadband connectivity, globally. The survey discusses emerging trends from the 5G techno-economic literature and makes five key recommendations for the design and standardization of Next Generation 6G wireless technologies.


  Access Paper or Ask Questions

Extending CLIP for Category-to-image Retrieval in E-commerce

Jan 04, 2022
Mariya Hendriksen, Maurits Bleeker, Svitlana Vakulenko, Nanne van Noord, Ernst Kuiper, Maarten de Rijke

E-commerce provides rich multimodal data that is barely leveraged in practice. One aspect of this data is a category tree that is being used in search and recommendation. However, in practice, during a user's session there is often a mismatch between a textual and a visual representation of a given category. Motivated by the problem, we introduce the task of category-to-image retrieval in e-commerce and propose a model for the task, CLIP-ITA. The model leverages information from multiple modalities (textual, visual, and attribute modality) to create product representations. We explore how adding information from multiple modalities (textual, visual, and attribute modality) impacts the model's performance. In particular, we observe that CLIP-ITA significantly outperforms a comparable model that leverages only the visual modality and a comparable model that leverages the visual and attribute modality.

* 15 pages, accepted as a full paper at ECIR 2022 

  Access Paper or Ask Questions

On some Foundational Aspects of Human-Centered Artificial Intelligence

Dec 29, 2021
Luciano Serafini, Raul Barbosa, Jasmin Grosinger, Luca Iocchi, Christian Napoli, Salvatore Rinzivillo, Jacques Robin, Alessandro Saffiotti, Teresa Scantamburlo, Peter Schueller, Paolo Traverso, Javier Vazquez-Salceda

The burgeoning of AI has prompted recommendations that AI techniques should be "human-centered". However, there is no clear definition of what is meant by Human Centered Artificial Intelligence, or for short, HCAI. This paper aims to improve this situation by addressing some foundational aspects of HCAI. To do so, we introduce the term HCAI agent to refer to any physical or software computational agent equipped with AI components and that interacts and/or collaborates with humans. This article identifies five main conceptual components that participate in an HCAI agent: Observations, Requirements, Actions, Explanations and Models. We see the notion of HCAI agent, together with its components and functions, as a way to bridge the technical and non-technical discussions on human-centered AI. In this paper, we focus our analysis on scenarios consisting of a single agent operating in dynamic environments in presence of humans.


  Access Paper or Ask Questions

Hyperparameter Optimization: Foundations, Algorithms, Best Practices and Open Challenges

Jul 14, 2021
Bernd Bischl, Martin Binder, Michel Lang, Tobias Pielok, Jakob Richter, Stefan Coors, Janek Thomas, Theresa Ullmann, Marc Becker, Anne-Laure Boulesteix, Difan Deng, Marius Lindauer

Most machine learning algorithms are configured by one or several hyperparameters that must be carefully chosen and often considerably impact performance. To avoid a time consuming and unreproducible manual trial-and-error process to find well-performing hyperparameter configurations, various automatic hyperparameter optimization (HPO) methods, e.g., based on resampling error estimation for supervised machine learning, can be employed. After introducing HPO from a general perspective, this paper reviews important HPO methods such as grid or random search, evolutionary algorithms, Bayesian optimization, Hyperband and racing. It gives practical recommendations regarding important choices to be made when conducting HPO, including the HPO algorithms themselves, performance evaluation, how to combine HPO with ML pipelines, runtime improvements, and parallelization.


  Access Paper or Ask Questions

<<
290
291
292
293
294
295
296
297
298
299
300
301
302
>>