51037
'Midnight Bus': A trundling bus trip to boresville | The Japan Times
by Contributing Writer Article history Online: Jan 24, 2018 Last Modified: Jan 24, 2018 Japanese filmmakers apparently can’t get enough of churning out paeans…
from instapaper
23 hours ago
GLUE排行榜上全面超越BERT的模型近日公布了!
机器之心 翻译 2019/02/13 14:28 Xiaodong Liu等 作者 在通用语言理解评估(GLUE) 基准 中,自 BERT 打破所有 11 项 NLP 的记录后,可应用于广泛任务的 NLP 预训练模型得到了大量关注。2018 年底, 机器之心 介绍了 微软 提交的综合性多任务 NLU 模型,它 在 11…
from instapaper
2 days ago
图解神经机器翻译中的注意力机制
几十年来, 统计 机器翻译 在翻译模型中一直占主导地位 [9],直到 神经 机器翻译 (NMT)出现。NMT 是一种新兴的 机器翻译 方法,它试图构建和训练单个大型 神经网络 ,该网络读取输入文本并输出译文 [1]。 NMT 的最初开拓性研究来自 Kalchbrenner 和 Blunsom…
from instapaper
2 days ago
简单的图神经网络介绍
最近,Graph Neural Network(GNN)在很多领域日益普及,包括社交网络、 知识图谱 、 推荐系统 甚至于生命科学。GNN在对节点关系建模方面表现十分突出,使得相关的研究领域取得了一定突破。本文旨在对GNN做一个简单的介绍,并介绍两种前沿算法,DeepWalk和GraphSage。 Graph…
from instapaper
2 days ago
Yale-LILY/LectureBank: LectureBank Dataset
LectureBank: a corpus for NLP Education and Prerequisite Chain Learning
This is the github page for our paper What Should I Learn First: Introducing LectureBank for NLP Education and Prerequisite Chain Learning in the proceedings AAAI 2019.
NLP 
2 days ago
Info 256. Applied Natural Language Processing
Applied Natural Language Processing
Info 256. Spring 2019
T/Th 12:30-2pm, 202 South Hall, UC Berkeley
David Bamman (office hours: Wednesday 10am-noon, 314 South Hall), dbamman@berkeley.edu
TA: Masha Belyi, mashabelyi@berkeley.edu
Course repo: https://github.com/dbamman/anlp19
course  nlp  deeplearning 
2 days ago
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
(Bidirectional Encoder Representations from Transformers)
Jacob Devlin
Google AI Language
nlp  deeplearning 
2 days ago
What is Linguistics?
Nathan Schneider
ENLP | 28 January 2019
NLP 
2 days ago
Twitter
RT : We've trained an unsupervised language model that can generate coherent paragraphs and perform rudimentary reading…
from twitter_favs
3 days ago
Better Language Models and Their Implications
Our model, called GPT-2 (a successor to GPT ), was trained simply to predict the next word in 40GB of Internet text. Due to our concerns about malicious…
from instapaper
3 days ago
WSDM19_FakeNews
Fake News
Fundamental Theories, Detection Strategies & Challenges
12th ACM International Conference on Web Search and Data Mining (WSDM), Tutorial

February 11, 2019 | Melbourne, Australia
NLP 
3 days ago
Why Do Smart People Send Nudes?
In so many ways, Jeff Bezos isn’t a relatable guy. As the CEO of Amazon, he’s the world’s richest man . He lords over much of the internet’s infrastructure . He…
from instapaper
4 days ago
Twitter
RT : Posted CMU "Neural Nets for NLP" lecture on sentence/contextualized word representations -- SkipThought, ELMo, BERT…
from twitter_favs
4 days ago
The Most Important Software Innovations
David A. Wheeler Revised 2017-09-07; First version 2001-08-01 Too many people confuse software innovations with other factors, such asthe increasing speed of…
from instapaper
5 days ago
Seymour Papert’s legacy: children, computers, and the future of programming education
Dong Liang Dec 13, 2016 My Lesson in BASIC In 1984, Deng Xiaoping visited Shanghai expo and made the famous statement “computer literacy needs to begin with…
from instapaper
6 days ago
Generating Sentences from a Continuous Space
Samuel R. Bowman∗
NLP Group and Dept. of Linguistics
Stanford University
sbowman@stanford.edu
Luke Vilnis∗
CICS
University of Massachusetts Amherst
luke@cs.umass.edu
Oriol Vinyals, Andrew M. Dai, Rafal Jozefowicz & Samy Bengio
Google Brain
{vinyals, adai, rafalj, bengio}@google.com
VAE 
7 days ago
艰难世道的“白日梦”哲学 专访导演张律
▲ 《庆州》是以张律导演的亲身经历为素材制作的影片。因此很多人问他“和主人公崔贤(朴海日饰)像不像”,每当这时张律都会回答说:“我和朴教授(白贤镇饰)更像。”朴教授一喝酒就会胡言乱语,但孔允熙(申敏儿饰)这样的美女握住他的手后便立刻变得温顺起来。照片=Invent Stone提供
from instapaper
8 days ago
Transparent AI for the Enterprise
Ajay Chander
Director, Digital Life Lab
Fujitsu Labs of America
XAI 
8 days ago
No thank you, Mr. Pecker – Jeff Bezos – Medium
RT : I’ve written a post about developments with the National Enquirer and its parent company, AMI. You can find it here:
from twitter_favs
9 days ago
Peering under the hood of fake-news detectors
New work from MIT researchers peers under the hood of an automated fake-news detection system, revealing how machine-learning models catch subtle but consistent…
from instapaper
10 days ago
Simplification
Simplification consists of modifying the content and structure of a text in order to make it easier to read and understand, while preserving its main idea and…
from instapaper
10 days ago
Is China’s corruption-busting AI system ‘Zero Trust’ being turned off for being too efficient?
What would you do if you had a machine to catch a thief? If you were a corrupt Chinese bureaucrat, you would want to ditch it, of course. Resistance by…
from instapaper
10 days ago
[1902.01718] End-to-End Open-Domain Question Answering with BERTserini
RT : BERTserini: combining the magic of BERT with Anserini for end-to-end open-domain QA:
from twitter_favs
10 days ago
Twitter
RT : BERTserini: combining the magic of BERT with Anserini for end-to-end open-domain QA:
from twitter_favs
10 days ago
Reading a neural network’s mind
Press Inquiries Share Media Resources 1 images for download Access Media Media can only be downloaded from the desktop version of this website. Share Leave a…
from instapaper
11 days ago
刘天昭: 我怀疑自己投入社会之初,就是为了安心离开它|刘天昭|生活|博客_新浪网
她曾经在做过五年的社论主笔,对于时代,她走进去又走出来,但是文学始终和她融在一起。 作者 | 武靖雅 区分作家的方式有很多种,比如可以用他们擅长的文体加以区分;比如可以用世代进行分隔;但还可以有一种区分方式,促成了我们接下来要涉猎的这样一群作家——…
from instapaper
11 days ago
阳光如此易逝,反而比在黑暗中更孤独
豆瓣电影 我看 影讯&购票 选电影 电视剧 排行榜 分类 影评 2018年度榜单 2018书影音报告 Mr Z 评论 雪莉:现实的愿景 4 2018-12-10 14:52:45…
from instapaper
11 days ago
Twitter
RT : We've developed a new approach to pretraining cross-lingual models for tasks. It significantly improves on the…
NLP  from twitter_favs
11 days ago
[1703.05028] Fonduer: Knowledge Base Construction from Richly Formatted Data
We focus on knowledge base construction (KBC) from richly formatted data. In contrast to KBC from text or tabular data, KBC from richly formatted data aims to extract relations conveyed jointly via textual, structural, tabular, and visual expressions. We introduce Fonduer, a machine-learning-based KBC system for richly formatted data. Fonduer presents a new data model that accounts for three challenging characteristics of richly formatted data: (1) prevalent document-level relations, (2) multimodality, and (3) data variety. Fonduer uses a new deep-learning model to automatically capture the representation (i.e., features) needed to learn how to extract relations from richly formatted data. Finally, Fonduer provides a new programming model that enables users to convert domain expertise, based on multiple modalities of information, to meaningful signals of supervision for training a KBC system. Fonduer-based KBC systems are in production for a range of use cases, including at a major online retailer. We compare Fonduer against state-of-the-art KBC approaches in four different domains. We show that Fonduer achieves an average improvement of 41 F1 points on the quality of the output knowledge base---and in some cases produces up to 1.87x the number of correct entries---compared to expert-curated public knowledge bases. We also conduct a user study to assess the usability of Fonduer's new programming model. We show that after using Fonduer for only 30 minutes, non-domain experts are able to design KBC systems that achieve on average 23 F1 points higher quality than traditional machine-learning-based KBC approaches.
nlp  knowledgegraph 
12 days ago
Ha Jin’s Self-Revealing Study of The Chinese Poet Li Bai
In his biography of Li Bai, the novelist Ha Jin narrates the banished poet’s unusual life, which, in some ways, mirrors the biographer’s. Photograph from Alamy…
from instapaper
12 days ago
Roomful of Teeth Is Revolutionizing Choral Music
Subscribe » Onward and Upward with the Arts February 11, 2019 From death metal to throat singing to alpine yodelling, the experimental group is changing what it…
from instapaper
12 days ago
[1510.03820] A Sensitivity Analysis of (and Practitioners' Guide to) Convolutional Neural Networks for Sentence Classification
Convolutional Neural Networks (CNNs) have recently achieved remarkably strong performance on the practically important task of sentence classification (kim 2014, kalchbrenner 2014, johnson 2014). However, these models require practitioners to specify an exact model architecture and set accompanying hyperparameters, including the filter region size, regularization parameters, and so on. It is currently unknown how sensitive model performance is to changes in these configurations for the task of sentence classification. We thus conduct a sensitivity analysis of one-layer CNNs to explore the effect of architecture components on model performance; our aim is to distinguish between important and comparatively inconsequential design decisions for sentence classification. We focus on one-layer CNNs (to the exclusion of more complex models) due to their comparative simplicity and strong empirical performance, which makes it a modern standard baseline method akin to Support Vector Machine (SVMs) and logistic regression. We derive practical advice from our extensive empirical results for those interested in getting the most out of CNNs for sentence classification in real world settings.
NLP 
12 days ago
Compensating for NLP’s Lack of Understanding – Towards Data Science
Better AI applications without AGI Katherine Bailey Feb 3 Photo by Patrick Tomasso on Unsplash The saying “a picture is worth a thousand words” does something…
from instapaper
13 days ago
Generalized Language Models
As a follow up of word embedding post, we will discuss the models on learning contextualized word vectors, as well as the new trend in large unsupervised…
NLP  from instapaper
14 days ago
Twitter
This is a super cool resource: Papers With Code now includes 950+ ML tasks, 500+ evaluation tables (including SOTA…
from twitter_favs
16 days ago
Twitter
We used architecture search to improve Transformer architecture. Key is to use evolution and seed initial populatio…
from twitter_favs
16 days ago
It is *not* possible to detect and block Chrome headless
By Evan Sangaline | January 18, 2018 Follow @sangaline 206 Star 111 A few months back, I wrote a popular article called Making Chrome Headless Undetectable in…
from instapaper
17 days ago
拉里·佩奇:一位有“知识漫游癖”的谷歌领袖
洛克希德·马丁公司(Lockheed Martin)核融合项目的工程师查尔斯·蔡斯(Charles Chase)回忆,三年前,他出席谷歌公司的Solve for X 大会,坐在白色的皮沙发上,这时,一个从来没见过的男人蹲下身来跟他说话。…
from instapaper
17 days ago
Weak Supervision: The New Programming Paradigm for Machine Learning
Weak Supervision: The New Programming Paradigm for Machine Learning Alex Ratner, Stephen Bach, Paroma Varma, Chris Ré And referencing work by many other members…
from instapaper
17 days ago
snorkel/tutorials/workshop at master · HazyResearch/snorkel
Summer School in Snorkel, Weak Supervision & Software 2.0
MachineLearning 
17 days ago
I Know You’ll Be Back: Interpretable New User Clustering and Churn Prediction on a Mobile Social Application
Carl Yang∗†
, Xiaolin Shi†
, Jie Luo†
, Jiawei Han∗
∗University of Illinois, Urbana Champaign, 201 N Goodwin Ave, Urbana, IL 61801, USA
†Snap Inc., 64 Market St, Venice, CA 90291, USA

{jiyang3, hanj}@illinois.edu, †
{xiaolin.shi, roger.luo}@snap.com
XAI 
17 days ago
Twitter
My implementation of Character Based ConvNets for text classification published by in 2015 is now…
from twitter_favs
17 days ago
Twitter
Kind of a stealth normalizing flow tutorial for NLP folks. Beginning to see these techniques used more in the area…
from twitter_favs
17 days ago
Twitter
Latent Flows for Discrete Sequences (): Experiments with (non-)autoregressive flows for disc…
from twitter_favs
17 days ago
Untitled (https://arxiv.org/pdf/1901.10548.pdf)
Latent Flows for Discrete Sequences (): Experiments with (non-)autoregressive flows for disc…
from twitter_favs
17 days ago
Twitter
RT : New research! We’re sharing a strong baseline for sentence embeddings that requires no training. Read about our fin…
from twitter_favs
17 days ago
Exploring random encoders for sentence classification - Facebook Code
WHAT THE RESEARCH IS: A strong, novel baseline for sentence embeddings that requires no training whatsoever. We explored various methods for computing sentence…
from instapaper
17 days ago
Twitter
Wow, all these years and I didn't know that you can actually paste PDF-exported equations back into LaTeXiT for fur…
from twitter_favs
17 days ago
为啥朝廷总抓不到俺——十年反党活动的安全经验汇总
好几天没上线,可能有读者以为俺出事了。别担心!俺21日还在回复评论,截止发这篇博文,【未】超出14天的期限,属于【正常静默】。 因为这篇博文要【全面地】分享俺十年反党活动的【技术】经验,牵涉到很多零碎的内容,整理起来多费了点精力和时间。 ★“朝廷想抓俺而不可得”正说明了——俺的防御措施基本靠谱…
from instapaper
18 days ago
Twitter
We are open-sourcing VeGANs, a small library to easily train various existing using .

You provide a…
GANs  from twitter_favs
18 days ago
Twitter
Out now: our new Python package. StanfordNLP provides native, neural (PyTorch) tokenization, POS tagging an…
NLProc  from twitter_favs
18 days ago
Twitter
You probably didn't know this: has a research division and their method for Named Entity Recognition is cu…
from twitter_favs
18 days ago
« earlier      
academic ai ajax algorithms api art async backbone.js bayesian bayesiannetworks bbs beauty beijing bigdata bioinformatics biology blog book bookmark business c++ censorship chatbot china circuits cloud clustering complexnetworks complexsystems conference courses crawler css culture database datamining datascience deeplearning democracy design development dictionary documentary download economy editor education elasticsearch em ember.js encoding english entertainment environment europe firefox flask flickr flight food france freedom functionalprogramming funny germany git golang google gps graph hadoop health history hongkong html html5 http hu_shih hust india informationextraction internet iphone ir iran ireland it japan java javascript jquery json justice keyphraseextraction knowledgegraph korea latex law lda learning liberalism life linux london love lucene lung_yingtai lyric machinelearning magazine map markdown marriage mathematicalexpression mathematics matrix mcmc media microsoft mongodb mooc mood movie mp3 murakami_haruki music neo4j news nlp nlproc node.js nodejs nosql novel nutch opinionmining optimization paper papers pca pda pdf perl photo photos pictures podcast poem poetry politics probability probabilitytheory programming python q&a recommendation restful rss ruby rubyonrails russia science search semanticweb sentiment sentimentanalysis sex shanghai smallworld socialmedia socialnetwork software solr spain spark sports statistics svd taiwan tdt telecom temp tibet tool travel tv twitter ubuntu uk unicode unlabeled us variationalinference video videos visa visualization web web2.0 windows wong_karwai word2vec wuhan xai 电影 龙应台

Copy this bookmark:



description:


tags: