다중혈관 관상동맥 환자에서 y-문합을 이용하여 양쪽 내흉동맥만을 사용한 우회술의 조기 성적

Data Mining - Concepts and Techniques.

Matrix Factorization Techniques for Recommender Systems

Click-through rate prediction is an essential task in industrial applications, such as online advertising. Recently deep learning based models have been proposed, which follow a similar Embedding&MLP paradigm. In these methods large scale sparse input features are first mapped into low dimensional embedding vectors, and then transformed into fixed-length vectors in a group-wise manner, finally concatenated together to fed into a multilayer perceptron (MLP) to learn the nonlinear relations among features. In this way, user features are compressed into a fixed-length representation vector, in regardless of what candidate ads are. The use of fixed-length vector will be a bottleneck, which brings difficulty for Embedding&MLP methods to capture user's diverse interests effectively from rich historical behaviors. In this paper, we propose a novel model: Deep Interest Network (DIN) which tackles this challenge by designing a local activation unit to adaptively learn the representation of user interests from historical behaviors with respect to a certain ad. This representation vector varies over different ads, improving the expressive ability of model greatly. Besides, we develop two techniques: mini-batch aware regularization and data adaptive activation function which can help training industrial deep networks with hundreds of millions of parameters. Experiments on two public datasets as well as an Alibaba real production dataset with over 2 billion samples demonstrate the effectiveness of proposed approaches, which achieve superior performance compared with state-of-the-art methods. DIN now has been successfully deployed in the online display advertising system in Alibaba, serving the main traffic.

Deep Interest Network for Click-Through Rate Prediction

Many interesting problems in machine learning are being revisited with new deep learning tools. For graph-based semi-supervised learning, a recent important development is graph convolutional networks (GCNs), which nicely integrate local vertex features and graph topology in the convolutional layers. Although the GCN model compares favorably with other state-of-the-art methods, its mechanisms are not clear and it still requires considerable amount of labeled data for validation and model selection. In this paper, we develop deeper insights into the GCN model and address its fundamental limits. First, we show that the graph convolution of the GCN model is actually a special form of Laplacian smoothing, which is the key reason why GCNs work, but it also brings potential concerns of over-smoothing with many convolutional layers. Second, to overcome the limits of the GCN model with shallow architectures, we propose both co-training and self-training approaches to train GCNs. Our approaches significantly improve GCNs in learning with very few labels, and exempt them from requiring additional labels for validation. Extensive experiments on benchmarks have verified our theory and proposals.

https://aaai.org/ocs/index.php/AAAI/AAAI18/paper/download/16098/16553

Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learning

Multi-types of user behavior data (e.g., clicking, adding to cart, and purchasing) are recorded in most real-world recommendation scenarios, which can help to learn users’ multi-faceted preferences. However, it is challenging to explore multi-behavior data due to the unbalanced data distribution and sparse target behavior, which lead to the inadequate modeling of high-order relations when treating multi-behavior data “as features” and gradient conflict in multi-task learning when treating multi-behavior data “as labels”. In this paper, we propose CIGF, a Compressed Interaction Graph based Framework, to overcome the above limitations. Specifically, we design a novel Compressed Interaction Graph Convolution Network (CIGCN) to model instance-level high-order relations explicitly. To alleviate the potential gradient conflict when treating multi-behavior data “as labels”, we propose a Multi-Expert with Separate Input (MESI) network with separate input on the top of CIGCN for multi-task learning. Comprehensive experiments on three large-scale real-world datasets demonstrate the superiority of CIGF.

/pdf/compressed-interaction-graph-based-framework-for-multi-22ua5mqa.pdf

Compressed Interaction Graph based Framework for Multi-behavior Recommendation

Learning embedding table plays a fundamental role in Click-through rate(CTR) prediction from the view of the model performance and memory usage. The embedding table is a two-dimensional tensor, with its axes indicating the number of feature values and the embedding dimension, respectively. To learn an efficient and effective embedding table, recent works either assign various embedding dimensions for feature fields and reduce the number of embeddings respectively or mask the embedding table parameters. However, all these existing works cannot get an optimal embedding table. On the one hand, various embedding dimensions still require a large amount of memory due to the vast number of features in the dataset. On the other hand, decreasing the number of embeddings usually suffers from performance degradation, which is intolerable in CTR prediction. Finally, pruning embedding parameters will lead to a sparse embedding table, which is hard to be deployed. To this end, we propose an optimal embedding table learning framework OptEmbed, which provides a practical and general method to find an optimal embedding table for various base CTR models. Specifically, we propose pruning the redundant embeddings regarding corresponding features' importance by learnable pruning thresholds. Furthermore, we consider assigning various embedding dimensions as one single candidate architecture. To efficiently search the optimal embedding dimensions, we design a uniform embedding dimension sampling scheme to equally train all candidate architectures, meaning architecture-related parameters and learnable thresholds are trained simultaneously in one supernet. We then propose an evolution search method based on the supernet to find the optimal embedding dimensions for each field. Experiments on public datasets show that OptEmbed can learn a compact embedding table which can further improve the model performance. 

OptEmbed

The invention relates to a recommendation method, a training method and device of a recommendation model and a storage medium. The method specifically comprises the steps that when a user does not have direct operation behavior data on an operation object, pseudo operation behavior data of the user on the operation object is used as a training sample for carrying out recommendation model trainingto obtain a recommendation model, and a to-be-recommended object recommended to the user is determined based on the obtained recommendation model. According to the invention, the user social information can be utilized to supplement the training sample data, and the data volume of the training sample data used for training the recommendation model is increased, so that the prediction accuracy of the object recommendation model obtained through training is effectively improved, and the accuracy of using the object recommendation model to perform object recommendation is improved. The method canbe applied to the field of intelligent recommendation in artificial intelligence (AI), object recommendation closer to user requirements is achieved by optimizing the prediction accuracy of the object recommendation model, and the accuracy of object recommendation can be effectively improved.

Recommendation method, training method and device of recommendation model and storage medium

A method for determining advertisement value, comprising: receiving advertisement information, the advertisement information comprising: an advertisement ID and operation information about the advertisement by a user (101); determining an advertisement value element according to an advertisement application platform, the advertisement value element being an element having an influence on the advertisement value (103); determining an advertisement value calculation strategy according to the advertisement value element (105); predicting a value of the advertisement value element according to the advertisement information to obtain an advertisement value element value (107); and obtaining, by means of calculation, the advertisement value of the advertisement by taking the advertisement value element value and the advertisement value calculation strategy as reference factors (109). A higher accuracy of advertisement value determination is realized by predicting an advertisement value element value, thereby making it possible to maximize revenue of an advertisement application platform.

Method and device for determining advertisement value

A probabilistic relational database is a probability distribution over a set of deterministic relational databases (namely, possible worlds). Efficient processing of updating information in probabilistic databases is required in several applications, such as sensor networking, data cleaning. As an important class of updating probabilistic databases, conditioning refines probability distribution of possible worlds, and possibly removing some of the possible worlds based on general knowledge, such as primary key constraints, functional dependencies and others. The existing methods for conditioning are exponential over the number of variables in the probabilistic database for an arbitrary constraint. In this paper, a constraint-based conditioning algorithm is proposed by only considering the variables in the given constraint without enumerating the truth values of all the variables in the formulae of tuples. Then we prove the correctness of the algorithm. The experimental study shows our proposed algorithm is more efficient comparing the work in the literatures.

Ruiming Tang

Papers

Compressed Interaction Graph based Framework for Multi-behavior Recommendation

OptEmbed

Recommendation method, training method and device of recommendation model and storage medium

Method and device for determining advertisement value

An Efficient Conditioning Method for Probabilistic Relational Databases