Memorizing All for Implicit Discourse Relation Recognition

Implicit discourse relation recognition is a challenging task due to the absence of the necessary informative clues from explicit connectives. An implicit discourse relation recognizer has to carefully tackle the semantic similarity of sentence pairs and the severe data sparsity issue. In this article, we learn token embeddings to encode the structure of a sentence from a dependency point of view in their representations and use them to initialize a baseline model to make it really strong. Then, we propose a novel memory component to tackle the data sparsity issue by allowing the model to master the entire training set, which helps in achieving further performance improvement. The memory mechanism adequately memorizes information by pairing representations and discourse relations of all training instances, thus filling the slot of the data-hungry issue in the current implicit discourse relation recognizer. The proposed memory component, if attached with any suitable baseline, can help in performance enhancement. The experiments show that our full model with memorizing the entire training data provides excellent results on PDTB and CDTB datasets, outperforming the baselines by a fair margin.

Performance enhancement of maximum ratio transmission in 5G system with multi-user multiple-input multiple-output

<span>The downlink multi-user precoding of the multiple-input multiple-output (MIMO) method includes optimal channel state information at the base station and a variety of linear precoding (LP) schemes. Maximum ratio transmission (MRT) is among the common precoding schemes but does not provide good performance with massive MIMO, such as high bit error rate (BER) and low throughput. The orthogonal frequency division multiplexing (OFDM) and precoding schemes used in 5G have a flaw in high-speed environments. Given that the Doppler effect induces frequency changes, orthogonality between OFDM subcarriers is disrupted and their throughput output is decreased and BER is decreased. This study focuses on solving this problem by improving the performance of a 5G system with MRT, specifically by using a new design that includes weighted overlap and add (WOLA) with MRT. The current research also compares the standard system MRT with OFDM with the proposed design (WOLA-MRT) to find the best performance on throughput and BER. Improved system results show outstanding performance enhancement over a standard system, and numerous improvements with massive MIMO, such as best BER and throughput. Its approximately 60% more throughput than the traditional systems. Lastly, the proposed system improves BER by approximately 2% compared with the traditional system.</span>