Key Phrases Flashcards
Operating memory vs storing device
Operating memory is internal, quick and low capacity
RAM- read access mem (lost after turning on)
ROM- read only memory (turn on comp)
Storing device external, slow and high capacity
System programs
End-user applications
Bios (basic input/output systemto getcomps started)
Operating systems
Drivers (software inte rface to hardware device)
Programming language
Last program used
operating system
set of software that controls the hardware of computer, the device, resource and common services for computer programs to function. (windows, UNIX, linux, android, ios, mac)
intranet vs internet
intranet- IPT internet protocol technology to share info and resources within an organization
internet- a global network of interconnected networks use standard internet protocol.
peer-to-peer architecture VS client-server architecture
peer to peer: network where computers are equally privileged, equipotent participants in app. ex: BitTorrent, skyoe, Bitcoin, I2P overlay
client-server: server is computer program provide services. client require service on server (cant change roles)
ranges of networks
PAN personal area network (bluetooth, cell phone, printer)
LAN local area network (one/few buildings)
MAN metropolitan area network (hospitals, cities)
WAN wide area network (internet)
descriptive VS inductive statistics
descriptive: summarize sample using median, mode, avg, standard dev
inductive: transfer fata of sample of population to whole population
incident VS prevalence VS mortality rate VS lethality of disease
incident is freq of disease
prevalence proportion of population have disease
mortality rate is rate of death in population
lethality is proportion of death among people sick w specific disease
inferential statistics
generalise properties of population from properties of sample (ex: test hypothesis +derive estimates)
sensitivity VS specificity
sensitivity is true positives (correct diagnose w disease)
specificity is true negative (diagnose correctly w out disease)
meta-analysis
quantitative method of combine result from studies (published) and making conclusions to evaluate effectiveness
brightness VS contrast
brightness- pixels
contrast- diff of brightness between 2 pixels
virus
spyware
malicious software
virus is malicous software (malfare) w small pieces of code
spyware- infiltration software that monitors users. to get info (pass, id)
malicious software brings harm to system. form of worm, virus, rojans, spyware, adware and rootkits. F: steal
image format and compression
formats of compression algorithms
resolution, image size (number of pixels) and colour depth (bits/pixel)
JPEG, PNG, BMP
cryptography
convert data into format that is unreadable
aliasing
stroboscopic effect
alliasing: causes diff signals to be distinguished (distortions)
stroboscopic effect: visual phenomenon caused by aliasing when motion represented by series of samples
frequency filtration
thresholding
remove f to suppress tettering signals +reduce background noise
method of image segmentation (grey scale to binary)
pixels VS voxel
pixel smallest ocntrollable element in picture=2D
(1 pixel =8 bit)
voxel is volumetric pixel =3D, used to visualize medical data
firewall
maintain security of private network
spam
phishing
pharming
unrequested electronic messages
fraudulent act to get private info by tricking email recipients
redirect website traffic to fake site
cracking vs hacking
cracking is just breaking in and hacking use this to destroy/steal data
patch vs service pack
patch inserting a code into program;temporary fixes
service pack updates to software version that fixes problem
Types of machine learning
Supervised: labeled data acts as supervisor. Learns acc to given data. Predict output
Unsupervised: no labeled data set. Tries to find patterns+ structures and makes clustered of similar data types
Reinforced: tries to find optimized sol. Reward+punishment based learning.
Artificial neural networks
Consists of neurons, synapses, weights, biases and functions. Has 3 layers
- input layer: taking input layer from external source and passes onto hidden layers of network. No computation.
- hidden layer: computation performed. Passes to output layer
- output layer: computes+ gives input to outside world
Deep neural network
Type of Ann w/ many neural network
- CNN: convulational network: computer vision + acoustic modeling for automatic speech recognition
- rnn recurrent neural network: data flows in any direction, language modeling or handwriting recognition
Computer vision cv, patching, graphics processing units GPU
Cv - file of computer science that mimics human vision often use CNN )
Patching - divide age slide (16GB) in smaller patches
Gpu- chip in graphics card to rapidly images
Natural language processing NLP
Understanding contents + language to correctly extract info.
Challenges: natural language understanding, speech recognition and natural language generation
Chat GPT
RLHF - reinforced learning from human feedback. using supervised
Classify, generate, segment = supervised
Search= unsupervised
PACS
Picturing archiving and communicating system- storing, retrieval and distribution of images
Integrate w’ hospital info system and radiology info system
DICOM
Digital imaging and communication in medicine
- standard for medical imaging
- network communications network
Sampling
quantization
Sampling is dividing image to pixels
Quantization is assign # to pixel which represents the color ( brightening) 8 bit quantization offers 28 = 256 pixel
Binary representation
Binary representation is coding the image in bites for writing into data file
Non Compressed/loose: raw BMP TIFF png
Compressedb losely : jpg, JPEG, LWF- wavelet transformation , GIF (8 bit)
Brightness
Grayscale images (1 bite for each pixel= 256 shades of grey)
Colour images- red channel or red or green is 256 bites
3 bites= 24 bites therefore truecolour image is 224
probability distribution
binomial- n of sucesses in yes/no w same probability of sucess
poisson- same but have diff success probability
gaussian-normal distribution