View Code? Open in Web Editor
NEW
This project forked from ecnudase/seminar-materials
组会ppt与论文--每一次的精心准备都值得留下记录😛
seminar-materials's Introduction
| Title |
Detail |
Author |
link |
| Improving Knowledge Graph Embedding Using Simple Constraints (ACL2018) |
将KG embedding为复数的形式，通过新增复数运算的约束来提高KG embedding的效果 |
Zhouxiaoxu |
ppt |
| Title |
Detail |
Author |
link |
| Knowledge-aware Attentive Neural Network for Ranking Question Answer Pairs (SIGIR2018) |
主要做的是问题答案的排序，利用知识图谱扩展句子的表示学习。 |
Liutingting |
ppt |
| Title |
Detail |
Author |
link |
| CoLink: An Unsupervised Framework for User Identity Linkage (AAAI2018) |
一个无监督的框架，用于进行不同网络之间的实体匹配(即找到结点之间的映射关系) |
Lina |
ppt |
| Title |
Detail |
Author |
link |
| A Walk-based Model on Entity Graphs for Relation Extraction (ACL2018) |
进行关系抽取时，考虑了任意两两实体对间的关系构成的网络 |
Kuangjun |
ppt |
| Title |
Detail |
Author |
link |
| Contextual String Embeddings for Sequence Labeling (COLING2018) |
（flair）预训练一个字符级别的语言模型，用其来对单词进行表征。通过将其与传统的word embedding进行拼接，提高了序列标注任务的性能 |
Chenyuanzhe |
ppt |
| Title |
Detail |
Author |
link |
| Question Difficulty Prediction for READING Problems in Standard Tests (AAAI2017) |
构造了TACNN框架来评估英语阅读问题难度，将文章、问题、选项作为特征输入，预测每个问题的难度系数 |
Zhurenyu |
ppt |
| Title |
Detail |
Author |
link |
| Learning from Semi-Supervised Weak-Label Data(AAAI2018) |
半监督弱标签下的多标签分类算法(数据集只有少部分标注，大部分未标注；数据标注的准确性无法保证) |
Fuyingnan |
ppt |
| Title |
Detail |
Author |
| Why should I trust you |
提出了开源工具"Lime"，能够解释样本的预测结果，并且增加模型本身的可解释性 |
Fuyingnan |
| Deep Residual Learning for Image Recognition |
提出了更深层次的卷积网络架构——残差网络，解决了传统模型中网络难以训练的问题 |
Zhurenyu |
| A Unified Probabilistic Framework for Name Disambiguation in Digital Library |
将姓名消歧问题formalize成一个隐马尔科夫随机场，并提出了参数估计的两阶段算法；提出了自动确定重名人数的auto K算法 |
Lina |
| JointExtractionofEntitiesandRelations |
将实体识别和关系提取统一为序列标注问题，使用同一个模型同时进行实体识别和关系提取 |
Kuangjun |
| Title |
Detail |
Author |
| Mask R-CNN |
提出了Mask R-CNN用于图像的实例分割 |
yuruonan |
| Deep Reinforcement Learning for Mention-Ranking |
采用神经网络和强化学习技术增加共指消解的准确率 |
chenyuanzhe |
| Question Answering with Subgraph Embeddings |
采用基于子图嵌入的方法，进行问答系统的训练和答案预测 |
tanglumin |
| RNN学习心得 |
介绍了RNN相关概念，讲解了梯度消失和权重冲突问题 |
yangkang |
| Title |
Detail |
Author |
| Reinforcement Learning for Relation Classification from Noisy Data |
提出一个新的关系分类模型，由实体选择器和关系分类器构成，能够在“Sentence Level”提取关系。将实体选择问题转换成强化学习问题。 |
GuHang |
| Pix2code: Generating Code from a Graphical User Interface Screenshot |
使用CNN和RNN的联合模型，将网页的UI图转化为对应的HTML代码 |
E Shen |
| JAVA GC机制 |
讲解了java的内存分配机制和垃圾回收机制 |
YinJiaLing |
| Title |
Detail |
Author |
| Convolutional Sequence to Sequence Learning |
An architecture based entirely on convolutional neural networks for sequence to sequence learning(such as NMT) |
CuiYiFeng |
| DeepFM：A Factorization-Machine based Neural Network for CTR Predicti |
回顾了过去的CTR模型，以及介绍了一系列基于深度学习的CTR模型（FNN,PNN,WDL） |
ChenLeiHui |
| Aspect Level Sentiment Classification with Deep Memory Network |
介绍了Memory Network，用于情感分析问题 |
Void-Yu |
| Title |
Detail |
Author |
| Learning Structured Representation for Text Classification via Reinforcement Learning |
使用ID-LSTM + HS-LSTM学习文本结构，并用策略梯度法进行强化学习 |
JinLiJiao |
| Human Action Adverb Recognition: ADHA Dataset And A Three-stream Hybrid Model |
贡献了一个数据集：x为人类动作的视频流序列，y为动作对应的副词。 例如识别接吻的视频是“甜蜜地”，"激动地"，“绅士地” ... |
SunChen |
| Title |
Detail |
Author |
| Deep Forest: Towards An Alternative to Deep Neural Network |
周志华提出的gcForest多粒度级联森林 |
FuYingNan |
| Structure Regularized Neural Network for Entity Relation Classification for Chinese Literature Text |
利用结构正则化简化句法结构，进行关系提取 |
KuangJun |
| Ranking-Based Name Matching for Author Disambiguation in Bibliographic Data |
KDD Cup 2013第二名，使用基于字符串和元路径的相似度进行作者姓名消歧 |
LiNa |
| Title |
Detail |
Author |
| Modeling Mention, Context and Entity with Neural Networks for Entity Disambiguation |
2015年实体消岐的最优模型，采用神经网络，使用了Embedding，卷积，神经张量网络等结构。 |
ChenYuanZhe |
| 解析HashTable，HashMap，ConcurrentHashMap |
讨论了java中该三种结构的特点，主要从多线程安全性、性能等方面考虑 |
YinJiaLing |
| Title |
Detail |
Author |
| Study about word embedding on sentiment subspace |
研究词向量中用于表达情感的向量子空间，目的是提高情感分类任务效果 |
Void-Yu |
| Title |
Detail |
Author |
| R-FCN: Object Detection via Region-based Fully Convolutional Networks |
目标检测网络。效果不比之前的RCNN，Fast-RCNN差，但是速度更快了。 |
God E |
| BiNE: Bipartite Network Embedding |
在二分图中采用了表示学习的方法，将节点embedding成向量，通过向量距离来度量节点的相似性。 训练过程类似Word2vec，使用了负采样，负样本的采样分布使用了LSH来代替频率 |
ChenLeiHui |
seminar-materials's People
Watchers
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent