With the omnipresence of deep learning approaches in modern signal processing, the generalization of models to unseen domains is of crucial importance. Various strategies have been proposed to address such unsupervised domain adaptation including reconstruction autoencoders and adversarial training. In this paper, we propose another strategy using teacher–student learning, where we train a student model for the target domain using pseudo-labels generated by a teacher model trained on the source domain. To improve the adaptation, we propose to exploit cross-version data, i.e., target-domain data without labels but existing in different versions, which are expected to share the same labels. We process the pseudo-labels from the teacher model by keeping only consistent labels or by interpolating conflicting labels. In this paper, we demonstrate the proposed strategy for a music transcription subtask (multi-pitch estimation) using different performances of the same composition. We compare this strategy with a baseline method using a reconstruction autoencoder. Furthermore, we combine our strategy with the baseline method by enforcing the cross-version consistency of the target domain predictions. Our results show that our strategy helps to improve the unsupervised domain adaptation from one instrument (piano) to other instruments (singing and orchestra). Especially for the difficult transfer from piano music to complex orchestral music, we obtain substantial improvements using the proposed approach.

Abstract. We prove two conjectures in spectral extremal graph theory involving the linear combinations of graph eigenvalues. Let [Formula: see text] be the largest eigenvalue of the adjacency matrix of a graph [Formula: see text] and [Formula: see text] be the complement of [Formula: see text]. A nice conjecture states that the graph on [Formula: see text] vertices maximizing [Formula: see text] is the join of a clique and an independent set with [Formula: see text] and [Formula: see text] (also [Formula: see text] and [Formula: see text] if [Formula: see text]) vertices, respectively. We resolve this conjecture for sufficiently large [Formula: see text] using analytic methods. Our second result concerns the [Formula: see text]-spread of a graph [Formula: see text], which is defined as the difference between the largest eigenvalue and least eigenvalue of the signless Laplacian of [Formula: see text]. It was conjectured by Cvetković, Rowlinson, and Simić [ Publ. Inst. Math., 81 (2007), pp. 11–27] that the unique [Formula: see text]-vertex connected graph of maximum [Formula: see text]-spread is the graph formed by adding a pendant edge to [Formula: see text]. We confirm this conjecture for sufficiently large [Formula: see text].

BACKGROUND: Drug-Drug interactions (DDIs) are a challenging problem in drug research. Drug combination therapy is an effective solution to treat diseases, but it can also cause serious side effects. Therefore, DDIs prediction is critical in pharmacology. Recently, researchers have been using deep learning techniques to predict DDIs. However, these methods only consider single information of the drug and have shortcomings in robustness and scalability. RESULTS: In this paper, we propose a multi-type feature fusion based on graph neural network model (MFFGNN) for DDI prediction, which can effectively fuse the topological information in molecular graphs, the interaction information between drugs and the local chemical context in SMILES sequences. In MFFGNN, to fully learn the topological information of drugs, we propose a novel feature extraction module to capture the global features for the molecular graph and the local features for each atom of the molecular graph. In addition, in the multi-type feature fusion module, we use the gating mechanism in each graph convolution layer to solve the over-smoothing problem during information delivery. We perform extensive experiments on multiple real datasets. The results show that MFFGNN outperforms some state-of-the-art models for DDI prediction. Moreover, the cross-dataset experiment results further show that MFFGNN has good generalization performance. CONCLUSIONS: Our proposed model can efficiently integrate the information from SMILES sequences, molecular graphs and drug-drug interaction networks. We find that a multi-type feature fusion model can accurately predict DDIs. It may contribute to discovering novel DDIs.

Research on automatic music transcription has largely focused on multi-pitch detection; there is limited discussion on how to obtain a machine- or human-readable score transcription. In this paper, we propose a method for joint multi-pitch detection and score transcription for polyphonic piano music. The outputs of our system include both a piano-roll representation (a descriptive transcription) and a symbolic musical notation (a prescriptive transcription). Unlike traditional methods that further convert MIDI transcriptions into musical scores, we use a multitask model combined with a Convolutional Recurrent Neural Network and Sequence-to-sequence models with attention mechanisms. We propose a Reshaped score representation that outperforms a LilyPond representation in terms of both prediction accuracy and time/memory resources, and compare different input audio spectrograms. We also create a new synthesized dataset for score transcription research. Experimental results show that the joint model outperforms a single-task model in score transcription.

Watermarking as a novel intellectual property (IP) protection technique can protect field-programmable gate array IPs from infringement. However, existing watermarking techniques may give away sensitive information during the public verification, which enables malicious verifiers or third parties to remove the embedded watermark and resell the design. Current zero-knowledge watermarking verification schemes can address the sensitive information leakage issue but are vulnerable to embedding attacks, which makes them ineffective in preventing the infringement denying of untrusted buyers (verifiers). This paper proposes a new publicly verifiable watermarking detection technique based on chaos-based zero-knowledge interaction and time stamping to resiliently resist the sensitive information leakage and embedding attacks, and is thus robust to the cheating from the prover, verifier, or third party. Experimental results and analysis show that the proposed method has better robustness than the most recent related literature.

This research retrieved surface net radiation using meteorological data and Landsat 5 TM images of the four seasons in the year 2009. Meanwhile the 65 different landscape metrics of each analysis unit were acquired using landscape analysis method. Then the most effective landscape metrics which could affect surface net radiation were determined by correlation analysis, stepwise regression, etc. The results showed that the spatial composition of land cover pattern has significant effect on surface net radiation. The proportions of bare land and forest land are effective and important factors which affect the changes of surface net radiation all the year round. But the spatial allocation of land cover pattern has no significant influence on surface net radiation. Moreover, the proportion of forest land is more capable of explaining surface net radiation than the proportion of bare land. This study is helpful in exploring the formation and evolution mechanism of urban heat island. It also gives theoretical hints and realistic guidance for urban planning and sustainable development.