2d:null 2e:["$","title",null,{"children":"SM3809_A11 | ychsu6 | Observable"}] 33:[{"id":"f5b21e2980b5c339dd73e912a74aa8bdd478df0768cc33d9bf4232663554e444d488dc84be7138aa1d4febaf35405b48ee7f11c5eb8e9aa3d7a91988c1c2b491","href":"https://static.observableusercontent.com/files/f5b21e2980b5c339dd73e912a74aa8bdd478df0768cc33d9bf4232663554e444d488dc84be7138aa1d4febaf35405b48ee7f11c5eb8e9aa3d7a91988c1c2b491","name":"Screenshot 2021-04-15 at 6.08.14 AM.png","type":"image/png","lastModified":1618438156742.247,"size":647612}] 31:{"id":"de57fd960bb94465","slug":"sm3809_a11","title":"SM3809_A11","thumbnail":"278e025bfaee64517d23f4d1b9a7c38cdff8b50988b5e170832a922307e558de","access_level":"public","create_time":"$D2021-04-14T21:03:45.886Z","update_time":"$D2021-04-21T22:48:50.334Z","view_time":null,"first_public_time":"$D2021-04-14T21:04:24.566Z","has_unpublished_edits":null,"fork":false,"stars":2,"starred":false,"owner":{"type":"team","id":"c3d91d7db2867b4d","login":"betahsu","name":"ychsu6","avatar_url":"https://avatars-next.observableusercontent.com/197b7792c36f626b9fe391a472486b8b6dc3cf3503a2bb52e365e7916c324a9e","tier":"starter_2024"},"version":184,"publish_version":184,"role":null,"authored":false,"authors":[{"type":"individual","id":"8205ffe96d046f8c","login":"betahsu","name":"ychsu6","avatar_url":"https://avatars-next.observableusercontent.com/197b7792c36f626b9fe391a472486b8b6dc3cf3503a2bb52e365e7916c324a9e","tier":null}],"members":[{"type":"individual","id":"8205ffe96d046f8c","login":"betahsu","name":"ychsu6","avatar_url":"https://avatars-next.observableusercontent.com/197b7792c36f626b9fe391a472486b8b6dc3cf3503a2bb52e365e7916c324a9e","tier":null,"role":"editor"}],"comments":[{"id":"dd8f381e871673f2","reply_to_id":null,"content":"this is strange -- I wonder if others experienced this","create_time":"$D2021-04-15T04:42:17.513Z","update_time":null,"resolved":false,"user":{"type":"individual","id":"32869872eb6f2f82","login":"dhowe","name":"Daniel Howe","avatar_url":"https://avatars-next.observableusercontent.com/a5081aaa56ae9990aa27642040dc61de3ce8cea67f6f778a5b864e41111ba7a3","tier":null}},{"id":"3533d7a5acc3e1a5","reply_to_id":"dd8f381e871673f2","content":"can you forward me the email?","create_time":"$D2021-04-15T04:50:59.084Z","update_time":null,"resolved":false,"user":{"type":"individual","id":"32869872eb6f2f82","login":"dhowe","name":"Daniel Howe","avatar_url":"https://avatars-next.observableusercontent.com/a5081aaa56ae9990aa27642040dc61de3ce8cea67f6f778a5b864e41111ba7a3","tier":null}}],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"title":"SM3809_A11","cells":[{"id":0,"value":"md`# SM3809_A11`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":4,"value":"md`\n## AI and Critical Practice`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":10,"value":"md`### READ\nReading: Arns, Inke's [*Read_Me, Run_Me, Execute_Me: Software and its Discontents...*](https://rednoise.org/readings/arns.pdf)\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":12,"value":"md`### REVIEW\n#### Research on ml5js - StyleTransfer\n##### Intro\n

Neural Style Transfer (NST) transfers the style of one image onto another one.\n

2 stages: A pre-trained feature extractor(train a model on one particular style) and a transfer network(apply this style to images.)\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":36,"value":"md`\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":94,"value":"md`##### Overview\n

\"Generative tool\" rather than \"Classification tool\" nor \"Pattern matching\".

Bg: Learning styleTransfer in traditional supervised learning approach is not pratical \n(need pairs of explicit input images, original images & artistic representations, which rarely exist)

NST use deep neural networks to help extract statistical features of images related to CONTENT and STYLE\n

Only a single style reference image is needed.

\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":170,"value":"md`\n##### Technique\n

\nTwo networks: a pre-trained feature extractor and a transfer network.\n

Pre-trained feature extractor:\n

Some layers learn to extract the content of an image (the shape of a dog or the position of a car), while others learn to focus on texture (the small brush strokes of a painter or the fractal patterns of nature). Style transfer exploits this by running two images through a pre-trained neural network, looking at the pre-trained network’s output at multiple layers, and comparing their similarity. Images that produce similar outputs at one layer of the pre-trained model likely have similar content, while matching outputs at another layer signals similar style.\n

Transfer network:\n

An image translation network that takes one image as input and outputs another image.\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":123,"value":"md``","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":121,"value":"md`##### Application\n

Deep Dream:\n

With the Combination of Neural Style Transfer and [*Feature detection- wiki #Detectors*](https://en.wikipedia.org/wiki/Feature_(computer_vision)), an interesting reversed preocedure is being explored. \n

The software was designed to detect faces and other patterns in images, with the aim of automatically classifying images. However, once trained, the network can also be run IN REVERSE, being asked to adjust the original image slightly so that certain given output contents (e.g. here are faces or certain animals) have higher confidence scores.\n

(This reversal procedure can also be used for visualizations to understand the emergent structure of the neural network better.)\n\nIn this example, the network has been trained to perceive dogs. Original image is applied with Deap Dream after 10 and 50 iterations.\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":134,"value":"md`\n\n\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":68,"value":"md`##### Other Application\nIt can be applied in: Photo and video editors, Commercial art, Gaming, Virtual Reality ... etc.\n

Example:\n

- [*deepart.io8*](https://deepart.io/) provides style images for users to upload photos, select style images, and create their own artworks.\n

- Immersive virtual worlds. [*Style transfer for a 360° VR video*](https://youtu.be/pkgMUfNeUCQ)\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":149,"value":"md`##### Prototyping/ possibly start of my final project\n(Not sure yet whether I would have the ability to create or do such research in these firleds, but I'm quite interested in these following questions/topics. But there are so many skills I should pick up before actually get into it...so yeah...🙈)\n

Iterations applied if the input pixels are adjusted?

To adjust weights or (further add on) input each iterations?

\n...\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":14,"value":"md`### SUBMIT\n#### My work submitted to the New River Journal: [*Human After All*](https://betahsu.github.io/test/) \n

Open the [*Twine + RiTa work*](https://betahsu.github.io/test/Human-After-All.html) in a new tab.\n

This work has changed a looooooot ;) Adjusted the whole story structre and the narrative sequence and flow have been more carefully arranged. \n

(Was submitted before the deadline but for some reason they replied via email saying that they are not able to accept the submission at this time...😭)\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W33","stdlib":"1","titling":"auto"}} 32:"$31:body"