Hijacker:{，} 

作品名稱：『 Hijacker:{，} 』

作品概述：

我們只能藉由自己的想像，去理解與勾勒對方的夢境。

〈Hijacker:{，}〉為2010年與荷蘭UMGC醫學中心（University of Medical Center Groningen）合作，觀察早產嬰兒腦部血液活動狀態而發想的創作。作品以「夢的照相機」為主要概念，運用機器學習（machine learning）重新轉換2010年因資料量判讀與其中結構高度複雜性而無法呈現的部分，延伸創造更多可能性。作品藉由賦予軟體程式特定的規則，讓腦波偵測機械真正具有「想像」的能力，藉由合成「夢境圖像」、做夢者腦波所產生的關鍵字資料，與做夢者記憶相互比較，讓第三方得以進一步探究他人的夢境，達到替做夢者想像出可能、或相似的夢境場景。

Title:『 Hijacker:{，} 』

Year: 2019

Material: EEG cap, EEG sensors, f/NIRS sensor, customized software, dataset

Size: size variable, frame size 50(H)cm * 50(W)cm

“We can only use our imagination to understand and sketch out other people's dreams.”

"Hijacker: {,}" is a concept that was conceived when we monitored blood activity in the brains of premature babies in cooperation with UMCG (University Medical Center Groningen) in 2010.

At that time, some parts could not be realized because the amount of data and the design of its structure were too complex. This time, machine learning has been brought into the work to redesign and simplify the data-collection process, making it easier and more flexible and creating more possibilities. 

By assigning certain rules to the software, the work can imagine possible or similar dream scenes for dreamers. Symbolically, the main concept is a “camera for dreams” that interprets the imagination given to the machine.

The work synthesizes dream images, and the keywords that the software generates from the brain-activity dataset are compared with the dreamer's own text description.

Basic procedure:

Dataset (Daily){set and categorized by image and keyword} ->

Dataset (sleep){collected based on EEG & NIRS} ->

Compare and re-categorize by machine learning -> pick up keywords <-> compose sentence by ML <-> parsing images and train new dataset<-> get image composition by cocoapi -> generate image.