Label propagation - Semisupervised Learning with Applications to NLPDavid Przybilla
?
Label propagation is a semi-supervised learning algorithm that propagates labels from a small set of labeled data points to unlabeled data points. The algorithm constructs a graph with nodes for each data point and weighted edges representing similarity between points. It then iteratively propagates the labels across the graph from labeled to unlabeled points until convergence, resulting in "soft" probabilistic labels for all points. The algorithm aims to minimize an energy function that encourages points connected by strong edges to receive similar labels. It performs well with limited labeled data by leveraging the graph structure to make predictions for unlabeled points.
This document provides notes for a math chapter on absolute value and reciprocal functions. It covers graphing and expressing absolute value functions as piecewise functions. Key points include: absolute value is defined as the distance from zero on the number line; absolute value functions are continuous and have the same x-intercept as the linear function inside the absolute value; taking the absolute value of a quadratic function preserves the vertex but changes the range. Care must be taken with the domains of absolute value functions depending on if the coefficient of the x-term inside is positive or negative.
Label propagation - Semisupervised Learning with Applications to NLPDavid Przybilla
?
Label propagation is a semi-supervised learning algorithm that propagates labels from a small set of labeled data points to unlabeled data points. The algorithm constructs a graph with nodes for each data point and weighted edges representing similarity between points. It then iteratively propagates the labels across the graph from labeled to unlabeled points until convergence, resulting in "soft" probabilistic labels for all points. The algorithm aims to minimize an energy function that encourages points connected by strong edges to receive similar labels. It performs well with limited labeled data by leveraging the graph structure to make predictions for unlabeled points.
This document provides notes for a math chapter on absolute value and reciprocal functions. It covers graphing and expressing absolute value functions as piecewise functions. Key points include: absolute value is defined as the distance from zero on the number line; absolute value functions are continuous and have the same x-intercept as the linear function inside the absolute value; taking the absolute value of a quadratic function preserves the vertex but changes the range. Care must be taken with the domains of absolute value functions depending on if the coefficient of the x-term inside is positive or negative.
2. Problem
? Kako za zadanu tablicu realizirati
sklop?
Iz zadane tablice mo?e se dobiti
funkcija i to :
?U obliku zbroja minterma (umno?aka)
?U obliku umno?ka maksterma (zbrojeva).
3. Realizacija minterma
i maksterma
? Minterm je operacija umno?ka
? Realizira se logi?kim I sklopom
? Za samo jednu kombinaciju vrijednosti varijabli
(ulaza) ima vrijednost 1
? Broj razli?itih minterma ovisi o broju varijabli (ulaza)
i iznosi 2n, n C broj varijabli (ulaza)
AB A B AB
0 0 0
0 1 1
Realiziraj preostale 1 0 0
minterme s 2 ulaza
1 1 0
4. Maksterm
? Maksterm je operacija zbroja
? Realizira se ILI logi?kim sklopom
? Za samo jednu kombinaciju svojih varijabli (ulaza) ima
vrijednost 0
? Broj razli?itih maksterma utvr?uje se na isti na?in kao
u slu?aju minterma
A+B A B A+B
0 0 1
0 1 1
Realiziraj preostale 1 0 1
maksterme s 2 ulaza 1 1 0
5. Realizacija sklopova
zbrojem minterma i umo?kom maksterma
? Od kombinacija varijabli za koje funkcija ima
vrijednost 1 dobije se zbroj minterma
? Minterm mora imati vrijednost 1 kada se u njega uvrsti
odgovaraju?a kombinacija vrijednosti varijabli
? Od kombinacija za koje funkcija ima
vrijednost 0 dobije se umno?ak maksterma
? Maksterm mora imati vrijednost 0 kada se u njega
uvrsti odgovaraju?a kombinacija vrijednosti varijabli
6. Primjer EX ILI funkcija
A B Y
0 0 0 A+ B Za A = B = 0 maksterm ima vrijednost 0
0 1 1 AB Za A = 0 i B = 1 minterm ima vrijednost 1
1 0 1 AB Za A = 1 i B = 0 minterm ima vrijednost 1
1 1 0 A+B Za A = B = 1 Maksterm ima vrijednost 0
Y = ( A + B) ? ( A + B)
Y = AB + A B
7. Realizacija
Y = AB + A B Y = ( A + B) ? ( A + B)
Zbroj minterma Umno?ak maksterma
8. EX ILI funkcija
? Rije? je o dva razli?ita oblika iste funkcije.
Y = ( A + B) ? ( A + B) =
= A A + A B + AB + BB =; A A = 0, BB = 0
= A B + AB
9. Logi?ki sklop isklju?ivo ILI
? Isklju?ivo (EX) ILI funkcija mo?e se
realizirati zbrojem minterma i umno?kom
maksterma
? Me?utim, postoji logi?ki sklop za tu funkciju
A B Y
0 0 0
0 1 1
1 0 1
Y = AB 1 1 0
Na izlazu ima vrijednost 1 ako je, isklju?ivo na jednom od
ulaza vrijednost 1.
10. Isklju?ivo (EX) NILI
A B Y
0 0 1
0 1 0
1 0 0
Y = AB 1 1 1
Funkcija:
Y = A B + AB
Y = ( A + B)( A + B)
11. Minimizacija
? Zakoni Booleove algebre primjenjuju se prilikom
minimizacije funkcije C algebarska metoda
? Minimizacija je postupak transformacije funkcije
tako da bude realizirana s najmanjim mogu?im
brojem logi?kih sklopova
? Algebarska metoda nije pouzdan na?in
minimizacije funkcija
? Postoje metode, poput primjene Karnaugh C
ovih tablica pomo?u kojih se funkcija pouzdano
minimizira
12. Primjer
? Realizirati sklop za zadanu funkciju:
Y = ( A + D) ? ABC + C + D ? B + AD
13. Minimizacija funkcije
Y = ( A + D) ? ABC + C + D ? B + AD Dvostruki komplement
= ( A + D)ABC + C ? D ? BAD De Morganovo pravilo
= ( A + D)ABC + CDB( A + D) De Morganovo pravilo
= A ABC + ABCD + ABCD + BCDD =0
= ABC( D + D) + BCD =1
= ABC + BCD Nacrtaj sklop
14. Pokus
? Program Logisim omogu?ava minimizaciju
(Pokus 3) upisane funkcije ili nacrtanog
sklopa
? Odabirom naredbe Analyze Circuit u
izborniku Project te kartice Minimized mo?e
se vidjeti kako bi se izvela minimizacija
putem Karnaug C ovih tablica
? Klikom na gumbe Set As Expression i Build
Circuit izvodi se minimizacija