"How much is my data worth?" a team from @ucbrise explores how to efficiently compute the Shapley Value, a classic concept from cooperative game theory, for data valuation https://t.co/N6XX5clpO6 pic.twitter.com/I4C0S212rZ
— Ben Lorica 罗瑞卡 (@bigdata) June 8, 2019
Data Validation for Machine Learning #SySML2019
— ML Review (@ml_review) June 8, 2019
Schema: used for validation, defined in ProtoBuf, automatic inference-based, use synthetic data generation to test model
Training-Serving Skew: interpretable distance metrics: max|pi − qi|, i∈S
et al.https://t.co/JHiGzFOFpG pic.twitter.com/zbt6y9BhQX
Butterfly Transform: An Efficient FFT Based Neural Architecture Design. https://t.co/DxBPsQsYs4 pic.twitter.com/pxibkdSoyq
— arxiv (@arxiv_org) June 8, 2019
A nice example of this: https://t.co/kPL5PPcLuQ
— hardmaru (@hardmaru) June 7, 2019
Nice new work from FAIR on assessing bias in datasets.
— Yann LeCun (@ylecun) June 7, 2019
Turns out image recognition systems will recognize everyday objects more reliably if the picture was shot in developed countries than... https://t.co/Nq2ErHND15
Very cool visualizations of different word senses being represented in later layers of BERT, by @_coenen, Emily Reif, Ann Yuan and collaborators.https://t.co/RiMjXseXxW pic.twitter.com/o6sKdPprAJ
— Chris Olah (@ch402) June 7, 2019
neural conversation modeling=next utterance prediction in learning, but neural conversation modeling≠next utterance prediction in inference. @uralik1, @jasonleeinf and i present multi-turn beam search to be less greedy across turns in neural chit-chat. https://t.co/rikv2ZthSx
— Kyunghyun Cho (@kchonyc) June 7, 2019
Object Detection systems (commercial and academic) are trained on biased data
— Soumith Chintala (@soumithchintala) June 7, 2019
This disproportionally affects accuracy in lower-income households and continents like Africa & Asia.
Work by my colleagues at FAIR using the Dollar Street dataset from GapMinderhttps://t.co/gdm7Ar9f4l pic.twitter.com/vmtkh3i4eo
Extremely excited to share the latest #sportsanalytics research with @CMU_Stats @FranciMatano @leerichardson09 Taylor Pospisil @stat_sam and @PittTweet @kpelechrinis Nick Granered - #GoingDeep: Continuous-time within-play valuation using #BigDataBowl data https://t.co/cTnIl5I6HM pic.twitter.com/iP6WfNlEyr
— Ron Yurko (@Stat_Ron) June 7, 2019
Come see our *best paper* today at *SEM at noon. This is work by @najoungkim that grew out of Ellie Pavlick's and my JSALT team project last summer. https://t.co/VXoSgzCKtQ #naacl2019 https://t.co/X5QYkbkZjv
— Sam Bowman (@sleepinyourhat) June 7, 2019
Non-#naacl2019 NLP news: Microsoft is near human performance on the standard evaluation for the https://t.co/ryDQeo2HU2 tasks—they've crossed our lower-bound estimate for human performance on the overall average score and on three individual tasks. pic.twitter.com/zKU7cTGkoU
— Sam Bowman (@sleepinyourhat) June 6, 2019
There were 46 papers submitted to the “dataset / software” category for #NeurIPS2019. 469 papers submitted to RL. Re-prioritization sorely needed in the field 🤦♀️🙅🏽♂️
— Eric Jang (@ericjang11) June 6, 2019
