🚨 The Uncanny Valley doesn't exist. 🚨

Mujahid Khalifah in his PhD research just showed that Japanese emotional speech models created from artificial voice datasets match original data in emotional expression.

What are the implications? 🧵

doi.org/10.1109/ACCESS… You do not need to collect huge high quality datasets for each emotion to produce a good emotional voice. Just use a raw voice model and fine-tune it for specific emotion(s) on small emotional data - even artificial data (!). The emotions will still be perfectly understandable.

9/28-29に北海道NLPミーティング/言語獲得と理解研究会を実施しました.釧路高専,北大(@HokkaidoUnivPR),北見工大(@KitamiOf)から15名の発表者が集まり,今北海道を代表する人工知能･自然言語処理の研究を発表しました.発表のまとめ,録画へのURLを以下の通り簡単にまとめました.🧵
#HokkaidoNLP #NLProc

https://twitter.com/mich_ptaszynski/status/1807693870886752676

@HokkaidoUnivPR @KitamiOf まず,全ての発表を北海道NLPのYouTubeチャンネルで公開しています.発表資料は動画の説明で確認してください.
YT Live:
Webex録画:
また,プログラムを北海道NLPのホームページで確認できます.

#HokkaidoNLP #NLProc #Hokkaidoyoutube.com/live/yuf3pz8z4…

sites.google.com/view/hokkaidon…

(っ◔◡◔)っ❤️ 𝗜𝘀 𝗶𝘁 𝗱𝗶𝗳𝗳𝗶𝗰𝘂𝗹𝘁 𝘁𝗼 𝘁𝗿𝗮𝗻𝘀𝗹𝗮𝘁𝗲 𝗶𝗿𝗼𝗻𝘆? 💔

"Initial Exploration into Sarcasm and Irony Through Machine Translation"
was just published in Natural Language Processing Journal @ElsevierConnect @sciencedirect

Check 🧵 for highlights and data With ChiaZhengLin, @mich_ptaszynski, @mar_kar_, @JuusoEronen, and Fumito Masui, we study if it's harder to translate irony compared to straightforward speech. We also specify the optimal conditions for finetuning transformers to translate irony, and compare with zero-shot GPT3.5.

⚡️Username toxicity paper alert⚡️
In our first huge-scale study (over 300 k users!) we found out that users with toxic usernames produced more toxic content and were more likely to have their account suspended. 😱
authors.elsevier.com/c/1fN212f~UWIs…
doi.org/10.1016/j.chb.… This is joint work with Rafal Urbaniak from @Uniwersytet_GD and the awesome team from @samurai_labs including @ptempska, @sandyire_, and @Fidofornia
#cyberbullyingawareness #cyberbullying #cyberbullyingprevention #Username #saferinternet