刘小邦的个人博客分享 http://blog.sciencenet.cn/u/iamliuzhiyong 浮生浪迹笑明月 千愁散尽一剑轻

博文

[机器学习]user interface engineering

已有 3978 次阅读 2013-9-18 19:59 |个人分类:机器学习|系统分类:教学心得| Interface

user interface engineering:

augumented reality in TMU and HCI in Univer M; Cambridge for Post Doc and HCI work; new group advanced interactive technology; sensor technology and interactive technology; all student from computer science; four doctor student; infomation system; two electronical engineering phd; some have visual computing class; and some have computer graphics;  some have machine learning course;

 

Lecture 1:

get you do exercise in class; design an implement; sensor user interface; semantic meaning of lower level stuff; sensor in first two weeks; working principal for selecting sensors; algorithm for interpret signal; buid working interactive system without;

computer: highly distributed; small farm factor; interaction: graphical user interface; harware user interface such as touch screen; multitouch hardware; input device;

power and system control; moor's law; small machine with higher and higer capacibility; iapd new form factor new user experience; most valuable skill intuitive interface; dvd recorder; 2 valve faucet? method and methodology; hci as discipline; including cognitive psychonology; computer science; electrical engineering; industry design. mechologdy from industry design.

design evluation and prototype; problem analysis; prototype; map to the requirement and interation; talk to people; unterstand; prototype; right size; right weight; for research; does the user interface better than other; exterlize; domian expert; low fidelity prototype;  functional visuable prototype; function trump form? relovationary? fundamental unsucessful running of battery and tow big; functional prototype; understanding design space; compacitive input; resister input; administrativa; next week homework assignment;  4 applied homework assignment; turn on their own assignment; 40% homework; 40% research project; 20%mid term; integrity; some of the implementation should don't look at open cv.

eth; computer history and hci history; mouse was roughly invented in 1963; very simple mechanical device; game changer; hci focus on demos; check it out; basic principle; monitor; keyboard; mouse; one finger click; senses; iphone; touch bechomes input; kinect; body configuration; define the next changes; touch; university of toronto design the touch suface; digital desk; digital calculator; smart kin give the procedure; ftir using camera and optical principle; xwand; perform motion gesture; kinect pose estimation pipeline using machine learning to infer body part per pixel; hypothessize body joint positon; indication of future; physical interaction; embracing 3d interface; phyically interaction in the mobile space; 3d reconstruction using kinect; sense gemoetry; holedesk; regular lcd screen; beam; kinect; in mobile space; weable controller; digits; consist of infrad camera; high fidelity reconstruction; steeable projection; augmented reality overview and debate for

Lecture 2

vison based; camera projected based; input recogniation; physical and digital world; physical sensors to sense the world; visual contact make abstract contact feasible; kinecmatci chain; bimanual interatation for writing; take the function of physical step; using camera and illumination led; low cost FTIR; Frustrated Total Inner Relfection; ration of angle; acyclical angle; serial leds along the edge; stretch photos; recover as much as finger informaton; recayst; from tracking to flow; you can do different things; electronic switchable diffusor; 3d manipulatin;


1, propagate from D to A if D is labeled, or else Retrieve B and C, if B is labeled, propagate from B to A, if C is labeled, propagate from C to A, if neigher B and C  

image segmentation and component analysis; project phase; ideas about project phases; centroid; two d touch example; filtering smoothing problem; associate one frame to another, statistics extract from data; principal component analysis; a set of observtion; independent island; propertis; correlated minor axes; moment against statistical tools; define zero; geometary moments; sum of the pixel in the region; raw moment; is not invariant to rotation; central moment; simplicity; computation over the centroid; major or minor axes; derived from the central moment, lenth is the eigenvalue of covariance matrix; x and y bounding box; ellipse fit; decide whether two blobs are overlaped or not; use intensivly but not super intuative; error is invisiable to; compensate; infer; sensor noise; looking at history or looking at; using noise; minor axes change quick a lot; accuracy and precision; repeat it over and over; moving average; merge sensor from different measurement; filter give better construction; slides over low pass filter; he is still alive ; pretty much every where; gps on CAR prediction and update the location; ship or rocket navigations; probabilistic inference; infer the state of the system; but can also  have observations; one measurement; several measurement; state is the minimum input for future state estimation; constant for transition matrix; influence matrix; one transition from one state to another state, recursive data processing system; white noise; initialize; one measurement; the bette measuement the better state we get; perfom another measuremtn/. weighted of the two measurement. kalman gain the variance ration; K is time dependent; it equals the variance rations; best prediction , and A motion model; B what your input are; H what part in the palance ; Q and R are sensor noise; experically derived; rule of thum; model precise, Kalman filtering introduction; predit new observaion and corriences, no control input; recompute the kalman giain; observable element; P AND Q covariance or variance; udate the kalman gain;  R 2.2 2.2 tunning parameters;

disparity; diffusion illuminator; camera acylic and led for FTIR; custom LED, custom led and circuit; two voltage wipers; slef capacitive or multal capacitive; robustness; external communication; good false positive; shunting; circurity analog or ditigal for precion suffer; shift touch point; latency; one of the key issues; IR relfective for pixelSense. conductive project;stroke; how object tracking behind;



https://blog.sciencenet.cn/blog-942948-725980.html

上一篇:[机器学习]object-oriented programming
下一篇:[机器学习]Seminar in Robotics, Systems, and Controls
收藏 IP: 195.176.110.*| 热度|

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2022-12-3 19:44

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部