Multi-Task Learning (MTL) has attracted increasing attention in recommender systems. A crucial challenge in MTL is to learn suitable shared parameters among tasks and to avoid negative transfer of information. The most recent sparse sharing models use independent parameter masks, which only activate useful parameters for a task, to choose the useful subnet for each task. However, as all the subnets are optimized in parallel for each task independently, it is faced with the problem of conflict between parameter gradient updates (i.e, parameter conflict problem). To address this challenge, we propose a novel Contrastive Sharing Recommendation model in MTL learning (CSRec). Each task in CSRec learns from the subnet by the independent parameter mask as in sparse sharing models, but a contrastive mask is carefully designed to evaluate the contribution of the parameter to a specific task. The conflict parameter will be optimized relying more on the task which is more impacted by the parameter. Besides, we adopt an alternating training strategy in CSRec, making it possible to self-adaptively update the conflict parameters by fair competitions. We conduct extensive experiments on three real-world large scale datasets, i.e., Tencent Kandian, Ali-CCP and Census-income, showing better effectiveness of our model over state-of-the-art methods for both offline and online MTL recommendation scenarios.

We present a novel robust control scheme that deals with multi-body spacecraft attitude tracking problems. The control scheme consists of a radial basis function network (RBFN) and a robust controller. By using the finite time convergence property of the terminal sliding mode (TSM), we derive a new online learning algorithm for updating all the parameters of the RBFN that ensures the RBFN has fast approximation for the parameter uncertainties and external disturbances. We design a robust controller to compensate RBFN approximation errors and realise the anticipative stability and performance properties. We can also achieve closed-loop system stability using Lyapunov stability theory. No detailed knowledge of the non-linear dynamics of the spacecraft is required at any point in the entire design process, and the proposed robust scheme is simple and effective and can be applied to more complex systems. Simulation results demonstrate the good tracking characteristics of the proposed control scheme in the presence of inertial uncertainties and external disturbances.

BACKGROUND: Miniature inverted repeat transposable elements (MITEs) are abundant non-autonomous elements, playing important roles in shaping gene and genome evolution. Their characteristic structural features are suitable for automated identification by computational approaches, however, de novo MITE discovery at genomic levels is still resource expensive. Efficient and accurate computational tools are desirable. Existing algorithms process every member of a MITE family, therefore a major portion of the computing task is redundant. RESULTS: In this study, redundant computing steps were analyzed and a novel algorithm emphasizing on the reduction of such redundant computing was implemented in MITE Digger. It completed processing the whole rice genome sequence database in ~15 hours and produced 332 MITE candidates with low false positive (1.8%) and false negative (0.9%) rates. MITE Digger was also tested for genome wide MITE discovery with four other genomes. CONCLUSIONS: MITE Digger is efficient and accurate for genome wide retrieval of MITEs. Its user friendly interface further facilitates genome wide analyses of MITEs on a routine basis. The MITE Digger program is available at: http://labs.csb.utoronto.ca/yang/MITEDigger.

Abstract Summary: TE Displayer can be used to retrieve genetic polymorphisms caused by transposable elements (TEs) in large-genomic datasets and present the results on virtual gel images. This enables researchers to compare TE profiles in silico and provides reference profiles for experimental analyses. Availability and implementation: Freely available on the web at http://labs.csb.utoronto.ca/yang/TE_Displayer. Contact: [email protected] Supplementary Information: Supplementary data are available at Bioinformatics online.

SUMMARY: Determination of transgene location is essential for investigating the effects of position on transgene expression levels and facilitates cloning of the resident gene affected by insertion. Currently used PCR-based approaches for determination of transgene location are relatively complicated and often fail when the transgene is duplicated, rearranged or fragmented. HideNseek is a new bioinformatics tool that allows computation of transgene locations, provided that a suitable genomic restriction enzyme digestion profile is available. Since the new approach is not based on the terminal sequences of the transgene insert, it is less sensitive to transgene duplication, rearrangement or fragmentation. HideNseek has been tested experimentally and by in silico simulation. The experimental example provided here shows that this simple approach is feasible, permitting rapid location of transgenes with little bench work. AVAILABILITY: available on request from the authors. SUPPLEMENTARY DATA: HideNseek input and output examples, experimental procedures and figures showing experimental results are provided as supplementary files: Supplementary material 1, 2, 3 and Supplementary figures (Figs 1 and 2), respectively. Supplementary data is available at Bioinformatics online.