Learning-based compression shows competitive compression ratios for genomics data. It often includes three types of compressors: static, adaptive and semi-adaptive. However, these existing compressors suffer from inferior compression ratios or throughput, and adaptive compressors also faces model cold-start problems. To address these issues, we propose DeepGeCo, a novel genomics data lossless adaptive compression framework with (s,k)-mer encoding and deep neural networks, involving three compression modes (MINI for static, PLUS for adaptive, ULTRA for semi-adaptive) for flexible requirements of compression ratios or throughput. In DeepGeCo, (1) we develop BiGRU and Transformer as the backbone to build Warm-Start and Supporter models in terms of cold-start problems. (2) We introduce (s,k)-mer encoding to pre-process genomics data before feeding it into the DNN model for improve model throughput, and we propose a new metric - Ranking of Throughput and Compression Ratio (RTCR) for effective encoding parameters selection. (3) We design a threshold controller and a probabilistic mixer within the backbone to balance compression ratios and model throughput. Experiments on 10 real-world datasets show that DeepGeCo's three compression modes improve up to a 22.949X average throughput and up to a 31.095% average compression ratio improvement while occupying low CPU or GPU memory.

Abstract As various types of data grow explosively, large-scale data storage, backup, and transmission become challenging, which motivates many researchers to propose efficient universal compression algorithms for multi-source data. In recent years, due to the emergence of hardware acceleration devices such as GPUs, TPUs, DPUs, and FPGAs, the performance bottleneck of neural networks (NN) has been overcome, making NN-based compression algorithms increasingly practical and popular. However, the research survey for the NN-based universal lossless compressors has not been conducted yet, and there is also a lack of unified evaluation metrics. To address the above problems, in this paper, we present a holistic survey as well as benchmark evaluations. Specifically, i) we thoroughly investigate NN-based lossless universal compression algorithms toward multi-source data and classify them into 3 types: static pre-training, adaptive, and semi-adaptive. ii) We unify 19 evaluation metrics to comprehensively assess the compression effect, resource consumption, and model performance of compressors. iii) We conduct experiments more than 4600 CPU/GPU hours to evaluate 17 state-of-the-art compressors on 28 real-world datasets across data types of text, images, videos, audio, etc. iv) We also summarize the strengths and drawbacks of NN-based lossless data compressors and discuss promising research directions. We summarize the results as the NN-based Lossless Compressors Benchmark (NNLCB, See fahaihi.github.io/NNLCB website), which will be updated and maintained continuously in the future.

Summary With the large‐scale popularity of mobile terminals, crowdsensing technology gradually replaces the existing static sensors with its advantages of high efficiency and low cost, becoming an emerging data collection method. How to assign perception tasks to the best performing users under the premise of ensuring quality and reducing costs to maximize the number of user tasks completed is the focus of the research on quantity sensitive task allocation. Based on this, a solution based on the improved whale optimization algorithm that combines the three operations of nonlinear decreasing convergence factor, optimal local jitter, and dynamic position update is put forward, which is used to solve the proposed task allocation problem. First, modeling the quantity sensitive task allocation problem, and then defining the spatial matching degree and skill matching degree according to the degree of adaptation between users and tasks. Taking into account the user's learning ability during the user's task execution, the skill update mechanism is introduced to update the user's existing skills in a timely manner, so as to improve task allocation effectiveness. Second, comprehensively considering the budget, the user's online time and the perceived task completion quality, and reasonably defining the task allocation problem that maximizes the number of tasks completed. In addition, from the perspective of selecting the best performing user for the task, designing a user selection strategy based on user's priority to reduce the cost of task allocation while ensuring the quality of the perceived task is basically completed. Then, in the process of solving the optimal task allocation plan, the improved algorithm is used to continuously optimize the initial task sequences of each iteration, and the final result can be obtained after a limited number of iterations. Finally, the improved algorithm is compared with other optimization algorithms in the same environment, and the results show that the improved algorithm has higher performance in solving task allocation problem.

With the development of mobile Internet, behavioral trajectories of human life are more and more recorded, which makes it possible to use computer technology to mine organizational behavior patterns. The mining of organizational behavior patterns based on social computing can not only prepare them in a targeted manner but also predict the consequences of possible measures. The organization behavior pattern mining has achieved a series of achievements in the fields of e-commerce and enterprise management. However, the problem of class imbalance and nonconsistent misclassification cost is common in the field of organizational behavior. For this problem, this article compares and analyzes the performance of the organizational behavior prediction model established by four typical cost-sensitive learning methods based on six classifiers, which provides a basis for the appropriate selection of cost-sensitive learning methods in different situations. Among them, the upsampling learning method is a better cost-sensitive learning method. However, there are some shortcomings in the upper sampling method. In order to avoid the possible overfitting problem of the social organization behavior prediction model established by the upper sampling method, this article proposes a new cost-sensitive learning method suitable for the mining of organizational behavior patterns. Based on the cost curve, this article proposes an effective personalized solution to the problem of class disequilibrium and nonconsistent misclassification cost in organizational behavior prediction modeling.