Neural network models have been demonstrated to be capable of achieving remarkable performance in sentence and document modeling. Convolutional neural (CNN) recurrent (RNN) are two mainstream architectures for such modeling tasks, which adopt totally different ways understanding natural languages. In this work, we combine the strengths both propose a novel unified model called C-LSTM representation text classification. utilizes CNN extract sequence higher-level phrase representations, fed...

Cloud computing, rapidly emerging as a new computation paradigm, provides agile and scalable resource access in utility-like fashion, especially for the processing of big data. An important open issue here is to efficiently move data, from different geographical locations over time, into cloud effective processing. The de facto approach hard drive shipping not flexible or secure. This work studies timely, cost-minimizing upload massive, dynamically-generated, geo-dispersed data cloud, using...

In edge-cloud computing, a set of servers (called edge servers) are deployed near the mobile devices to allow these offload their jobs and subsequently obtain results from with low latency. One fundamental problem in systems is how dispatch schedule so that job response time (defined as interval between release arrival computation result at device) minimized. this paper, we propose general model for problem, where generated arbitrary order times then offloaded both upload download delays....

Chinese calligraphy is among the finest and most important of all art forms an inseparable part history. Its delicate aesthetic effects are generally considered to be unique calligraphic arts. subtle power integral traditional painting. A novel intelligent system uses a constraint-based analogous-reasoning process automatically generate original that meets visually requirements. We propose can create novel, aesthetically appealing from few training examples existing styles. To demonstrate...

Network Function Virtualization (NFV) is a promising technology that promises to significantly reduce the operational costs of network services by deploying virtualized functions (VNFs) commodity servers in place dedicated hardware middleboxes. The VNFs are typically running on virtual machine instances cloud infrastructure, where virtualization enables dynamic provisioning VNF instances, process fluctuating traffic needs go through service. In this paper, we target enterprise - expressed as...

Efficient virtual machine (VM) management can dramatically reduce energy consumption in data centers. Existing VM algorithms fall into two categories based on whether the VMs' resource demands are assumed to be static or dynamic. The former category fails maximize utilization as they cannot adapt dynamic nature of demands. Most approaches latter heuristical and lack theoretical performance guarantees. In this work, we formulate a large-scale Markov Decision Process (MDP) problem derive an...

A distributed Java Virtual Machine (DJVM) spanning multiple cluster nodes can provide a true parallel execution environment for multi-threaded applications. Most existing DJVMs suffer from the slow in interpretive mode and thus may not be efficient enough solving computation-intensive problems. We present JESSICA2, new DJVM running JIT compilation that execute applications transparently on clusters. JESSICA2 provides single system image (SSI) illusion to via an embedded global object space...

We propose a new recommendation algorithm for online documents, images and videos, which is personalized. Our idea to rely on the attention time of individual users captured through commodity eye-tracking as essential clue. The prediction user interest over certain item (a document, image or video) based user's acquired using vision-based during his previous reading, browsing video watching sessions same type materials. After acquiring times collection materials, our can predict probable...

The rapidly increasing power of personal mobile devices (smartphones, tablets, etc.) is providing much richer contents and social interactions to users on the move. This trend however throttled by limited battery lifetime unstable wireless connectivity, making highest possible quality service experienced not feasible. recent cloud computing technology, with its rich resources compensate for limitations connections, can potentially provide an ideal platform support desired services. Tough...

We introduce several novel word features for keyword extraction and headline generation. These new are derived according to the background knowledge of a document as supplied by Wikipedia. Given document, acquire its from Wikipedia, we first generate query searching Wikipedia corpus based on key facts present in document. then use find articles that closely related contents With search result article set, extract inlink, outlink, category infobox information each derive set which reflect...

We propose a new document summarization algorithm which is personalized. The key idea to rely on the attention (reading) time of individual users spent single words in as essential clue. prediction user over every word based user's during his previous reads, acquired via vision-based commodity eye-tracking mechanism. Once attentions small collection are known, our can predict through semantics analysis. Our then summarizes according document. With algorithm, we have developed prototype...

A coflow is a collection of related parallel flows that occur typically between two stages multi-stage computing task in network, such as shuffle MapReduce. The abstraction allows applications to convey their semantics the network so application-level requirements can be better satisfied. In this paper, we study routing and scheduling multiple coflows minimize total weighted completion time (CCT). We first propose rounding-based randomized approximation algorithm, called OneCoflow, for...

We present a novel reinforcement learning architecture that learns structured representation for use in symbolic melody harmonization. Probabilistic models are predominant harmonization tasks, most of which only treat notes as independent observations and do not take note substructures the melodic sequence. To fill this gap, we add substructure discovery crucial step automatic chord generation. The proposed method consists module generates hierarchical structures melodies, policy to break...

An automatic algorithm can generate Chinese calligraphy by quantitatively representing the characteristics of personal handwriting acquired from learning examples.

Efficient virtual machine (VM) management can dramatically reduce energy consumption in data centers. Existing VM algorithms fall into two categories based on whether the VMs' resource demands are assumed to be static or dynamic. The former category fails maximize utilization as they cannot adapt dynamic nature of demands. Most approaches latter heuristic and lack theoretical performance guarantees. In this paper, we formulate a large-scale Markov decision process (MDP) problem derive an...

Target-specific stance detection on social media, which aims at classifying a textual data instance such as post or comment into class of target issue, is an emerging opinion mining paradigm importance. An example application would be to overcome vaccine hesitancy in combating the coronavirus pandemic. Existing strategies rely merely individual instances cannot always capture expressed given target. We address new task called conversational (CSD) infer toward (e.g., COVID-19 vaccination)...

We present an adaptive fault-tolerant wormhole routing algorithm for 2D meshes. The main feature is that with the algorithm, a normal message, when blocked by some faulty processes would detour along f-polygons around fault region. proposed can tolerate convex faults only three virtual channels per physical channel regardless of overlapping different regions. deadlock-free.

Training a computer to evaluate the aesthetics of Chinese characters provides feedback mechanism improve quality automatically generated calligraphy.

Auction design has recently been studied for dynamic resource bundling and VM provisioning in IaaS clouds, but is mostly restricted to the one-shot or offline setting. This work targets a more realistic case of online auction design, where: (i) cloud users bid resources into future assemble customized VMs with desired occupation durations; (ii) provider dynamically packs multiple types on heterogeneous physical machines (servers) requested VMs; (iii) operational costs servers are considered...