Cross-domain collaborative filtering (CDCF) is an effective method to alleviate the data sparsity problem by transferring knowledge from a source domain to assist the learning of a target domain. However, most of the existing CDCF approaches require that the two domains have at least one overlapping side (either on user or item) and the raw data can be fully shared across domains, which is difficult to be satisfied in reality due to corporate barriers and the risk of user privacy leakage. Although there are some attempts on applying CDCF to the scenario without overlapping data by transferring cluster-level rating patterns, these methods fail to mine the complex connections between the two domains, which makes their performance still not satisfactory. To address these problems, we propose a novel deep Interaction Distribution Transfer (IDT) framework, which extracts and transfers knowledge from the feature distribution formed by the whole dataset rather than specific data. In this way, the knowledge is embedded into high-order features for transfer, which can effectively avoid privacy leakage during the data sharing process. Moreover, as a flexible framework, IDT obtains powerful feature extraction ability from the base model, which guarantees its superior performance. Extensive experiments on three benchmark datasets are conducted and the results verify the effectiveness of the proposed framework.

Reinforcement learning-based portfolio management has recently attracted extensive attention. However, deep reinforcement learning methods are unexplainable and considered to be potentially risky, difficult to be trusted and regulated by users. To address these problems, we propose an eXplainable reinforcement learning framework for Portfolio Management, named XPM, which is efficient, concise, and can provide faithful explanations for network outputs. Specifically, we first design a policy network for portfolio management, which uses temporal convolutional network (TCN) to extract temporal features of multiple time series in portfolio. Then, we employ global average pooling (GAP) and a fully connected layer to integrate the global feature maps to handle asset correlations. Finally, we utilize softmax to determine the output portfolio weights. To assemble explainability into our model, we employ an explainable artificial intelligence method, class activation mapping (CAM), to explain the network outputs, which computes an activation map for an asset of interest. The map highlights the important assets and time intervals in the input state. In this way, end users can understand which part of the portfolio's recent price movements makes the network decision to invest in the target asset. Experimental results show that XPM outperforms the current state-of-the-art portfolio management methods in NASDAQ and NYSE markets, and can provide faithful and informative explanations to end users.

Data sparsity is a challenge problem that most modern recommender systems are confronted with. By leveraging the knowledge from relevant domains, the cross-domain recommendation technique can be an effective way of alleviating the data sparsity problem. In this paper, we propose a novel Bi-directional Transfer learning method for cross-domain recommendation by using Graph Collaborative Filtering network as the base model (BiTGCF). BiTGCF not only exploits the high-order connectivity in user-item graph of single domain through a novel feature propagation layer, but also realizes the two-way transfer of knowledge across two domains by using the common user as the bridge. Moreover, distinct from previous cross-domain collaborative filtering methods, BiTGCF fuses users' common features and domain-specific features during transfer. Experimental results on four couple benchmark datasets verify the effectiveness of BiTGCF over state-of-the-art models in terms of bi-directional cross domain recommendation.

Due to its nature of learning from dynamic interactions and planning for long-run performance, reinforcement learning (RL) recently has received much attention in interactive recommender systems (IRSs). IRSs usually face the large discrete action space problem, which makes most of the existing RL-based recommendation methods inefficient. Moreover, data sparsity is another challenging problem that most IRSs are confronted with. While the textual information like reviews and descriptions is less sensitive to sparsity, existing RL-based recommendation methods either neglect or are not suitable for incorporating textual information. To address these two problems, in this paper, we propose TDDPG-Rec, a Text-based Deep Deterministic Policy Gradient framework for interactive recommendation. Specifically, we leverage textual information to map items and users into a feature space, which greatly alleviates the sparsity problem. Moreover, we design an effective method to construct an action candidate set. By the policy vector dynamically learned from TDDPG-Rec that expresses the user's preference, we can select actions from the candidate set effectively. Through extensive experiments on three public datasets, we demonstrate that TDDPG-Rec achieves state-of-the-art performance over several baselines in a time-efficient manner.

Financial portfolio management is the process of periodically reallocating a fund into different financial investment products, with the goal of achieving the maximum profits. While conventional financial machine learning methods try to predict the price trends, reinforcement learning based portfolio management methods makes trading decisions according to the price changes directly. However, existing reinforcement learning based methods are limited in extracting the price change information at single-scale level, which makes their performance still not satisfactory. In this paper, inspired by the Inception network that has achieved great success in computer vision and can extract multi-scale features simultaneously, we propose a novel Ensemble of Identical Independent Inception (EI$^3$) convolutional neural network, with the objective of addressing the limitation of existing reinforcement learning based portfolio management methods. With EI$^3$, multiple assets can be processed independently while sharing the same network parameters. Moreover, price movement information for each product can be extracted at multiple scales via wide network and then aggregated to make trading decision. Based on EI$^3$, we further propose a recurrent reinforcement learning framework to provide a deep machine learning solution for the portfolio management problem. Comprehensive experiments on the cryptocurrency datasets demonstrate the superiority of our method over existing competitors, in both upswing and downswing environments.

With the advance of internet, multi-modal data can be easily collected from many social websites such as Wikipedia, Flickr, YouTube, etc. Images shared on the web are usually associated with social tags or other textual information. Although existing multi-modal methods can make use of associated text to improve image annotation, the disadvantages of them are that associated text is also required for a new image to be predicted. In this paper, we propose the cross-modal multi-task learning (CMMTL) framework for image annotation. Labeled and unlabeled multi-modal data are both levaraged for training in CMMTL, and it finally obtains visual classifiers which can predict concepts for a single image without any associated information. CMMTL integrates graph learning, multi-task learning and cross-modal learning into a joint framework, where a shared subspace is learned to preserve both cross-modal correlation and concept correlation. The optimal solution of the proposed framework can be obtained by solving a generalized eigenvalue problem. We conduct comprehensive experiments on two real world image datasets: MIR Flickr and NUS-WIDE, to evaluate the performance of the proposed framework. Experimental results demonstrate that CMMTL obtains a significant improvement over several representative methods for cross-modal image annotation.

A large number of practical domains, such as scene classification and object recognition, have involved more than two classes. Therefore, how to directly conduct multiclass classification is being an important problem. Although some multiclass boosting methods have been proposed to deal with the problem, the combinations of weak learners are confined to linear operation, namely weighted sum. In this paper, we present a novel large-margin loss function to directly design multiclass classifier. The resulting risk, which guarantees Bayes consistency and global optimization, is minimized by gradient descent or Newton method in a multidimensional functional space. At every iteration, the proposed boosting algorithm adds the best weak learner to the current ensemble according to the corresponding operation that can be sum or Hadamard product. This process grown in an adaptive manner can create the sum of Hadamard products of weak learners, leading to a sophisticated nonlinear combination. Extensive experiments on a number of UCI datasets show that the performance of our method consistently outperforms those of previous multiclass boosting approaches for classification.

Automatic image annotation is an important task for multimedia retrieval. By allocating relevant words to un-annotated images, these images can be retrieved in response to textual queries. There are many researches on the problem of image annotation and most of them construct models based on joint probability or posterior probabilities of words. In this paper we estimate the probabilities that words generate the images, and propose a two-phase generation model for the generation procedure. Each word first generates its related words, then these words generate an un-annotated image, and the relation between the words and the un-annotated image is obtained by the probability of the two-phase generation. The textual words usually contain more semantic information than visual content of images, thus the probabilities that words generate images is more reliable than the probability that images generate words. As a result, our model estimates the more reliable probability than other probabilistic methods for image annotation. The other advantage of our model is the relation of words is taken into consideration. The experimental results on Corel 5K and MIR Flickr demonstrate that our model performs better than other previous methods. And two-phase generation which considering word's relation for annotation is better than one-phase generation which only consider the relation between words and images. Moreover, the methods which estimate the generative probability obtain better performance than SVM which estimates the posterior probability.

In this paper, a semantic model for cross-modal and multi-modal retrieval is studied. We assume that the semantic correlation of multimedia data from different modalities can be depicted in a probabilistic generation framework. Media data from different modalities can be generated by the same semantic concepts, and the generation process of each media data is conditional independent under the semantic concepts. The semantic generation model (SGM) for cross-modal and multi-modal analysis is proposed based on this assumption. We study two types of methods: direct method Gaussian distribution and indirect method random forest, to estimate the semantic conditional distribution of SGM. Then methods for cross-modal and multi-modal retrieval are derived from SGM. Experimental results show that SGM based methods for cross-modal retrieval improve the accuracy over the state-of-the-art cross-modal method, but don't increase the time consuming, and the SGM multimodal retrieval methods also outperform traditional methods in image retrieval. Moreover, indirect SGM based method outperforms direct SGM method in the two types of retrieval, which proves that indirect SGM can better describe the semantic distribution.