0

For smoker,,I think, this is permitted-halal hehe

Posted by dayad on 22.55
Study: E-Cigarettes Fail at Nicotine Delivery, No Better Than Unlit Cigarette
Jason Mick (Blog) - February 10, 2010 11:42 AM


Despite popularity, the questions about e-cigarettes may not be all they are cracked up to be


Last year we wrote on the health risks associated with electronic cigarettes, commonly known as "e-cigarettes". The devices have been billed as "healthy living" products and as a tool to help smokers quit their addiction. Advocates say that since electronic cigarettes simply give smokers a vapor with nicotine and no burned chemicals, that they are relatively safe.

Those claims may be inaccurate, though. Last March, the Food and Drug Administration banned imports of the devices, which are largely manufactured in China. The FDA wants to investigate health concerns. Namely, the FDA found that chemical formulas for the smoky vapor often contained dangerous components; at least one manufacturer used diethylene glycol as a key ingredient, a chemical commonly used in antifreeze and toxic to humans.

Now a new study adds to the doubts about e-cigarettes, indicating that they are about as successful at delivering nicotine as puffing on an unlit cigarette. Dr. Thomas Eissenberg at the Virginia Commonwealth University led the study. The study involved 16 participants and extensively monitored nicotine levels in the body and heart rates when using both traditional and electronic cigarettes.

The study, the first study of e-cigarettes to be conducted by U.S. doctors, found that almost no nicotine was actually delivered by the devices and instead users were actually inhaling a nicotine-devoid toxic vapor of compounds like diethylene glycol or nitrosamines, a family of cancer-causing nitrogen compounds.

Describes Dr. Eissenberg, "They are as effective at nicotine delivery as puffing on an unlit cigarette. These e-cigs do not deliver nicotine. Ten puffs from either of these electronic cigarettes with a 16 mg nicotine cartridge delivered little to no nicotine."

The study was funded by the National Cancer Institute and will soon be published in the journal Tobacco Control, a product of the British Medical Journal Group.

Nicotine has some beneficial health effects, particularly for the mentally ill, so it is disappointing that e-cigarettes appear unable to deliver the compound.

Despite the mounting criticisms, many e-cigarette users stand by the product. Jimi Jackson, a former tobacco smoker in Richmond, Virginia, who sells electronic cigarettes, comments, "I smoked 37 years, and when I found them, I was, like, 'Thank, you Jesus.'"

The FDA is currently being sued by a company called "Smoking Everywhere" that imports e-cigarettes from China. The company wants the FDA to lift the ban on e-cigarette imports. The company's court filings reveal just how popular the devices are -- the company sold 600,000 e-cigarettes in a year via the company's network of 120 distributors in the United States.

Why should the FDA lift its ban? According to Washington lawyer Kip Schwartz, representing "Smoking Everywhere", "We are on the verge of going out of business, which is why we are suing the FDA in U.S. District Court."

source : www.dailytech.com

2

belajar JST lagi,,hehehe

Posted by dayad on 21.57 in
Jaringan Syaraf Tiruan adalah suatu cara pembelajaran dengan meniru cara kecerdasan manusia. Tapi yang harus di bawahi adalah kemampuan belajar jaringan syaraf tiruan adalah terbatas, jadi tidak bisa melakukan segalanya.
Proses pembelajaran jaringan syaraf tiruan terbagi menjadi dua, encode (menyimpan pengetahuan) dan decode (menanggapi dan memproses data yang masuk). Encode sendiri juga terbagi menjadi dua, supervised learning (pembelajaran yang dibimbing) dan unsupervised learning (tidak dibimbing). Sedangkan decode juga terbagi dua, feedforward dan feedback. Dalam konsep supervised learning masukan dan keluaran ditentukan oleh pembuat atau pengajarnya, dalam proses tersebut JST akan menyesuaikan prosesnya...bersambung hehe,,ngantuk

2

Strange is The Song in Our Conversation - Monkey To Millionaire

Posted by dayad on 15.39
Walking on the edge of an empty pool
That we thought it was cool
We lower down our voices
We're holding our laughs
Cos the moon are standing there without a noise

You're asking 'bout a boyfriend and girlfriend thing
Into our late night conversation
You sounded like you're so interested, so I feel like
I want to share it with you

After few hours of our conversation
You told me, you only sleep when there's someone singin'
So I sing to you a song called a "STRANGE"

Then I heard no more words
Coming out from you
Then I heard no more words
Coming out from you
"Strange" is the song in our conversation
"Strange" is the song in our conversation

Couple of stories has already been told
and few things about has already been told
You're sound so quiet so I whisper to you
Well I guess that you have found your own world yeah

Couple of stories has already been told
and few things about has already been told
You're sound so quiet so I whisper to you
Well I guess it's time to end it so I said to you
"I hope that we could talk again someday"

After few hours of our conversation
You told me, you only sleep when there's someone singin'
So I sing to you a song called a "STRANGE"

Then I heard no more words
Coming out from you
Then I heard no more words
Coming out from you
"Strange" is the song in our conversation
"Strange" is the song in our conversation

After few hours of our conversation
You told me, you only sleep when there's someone singin'
So I sing to you a song called a "STRANGE"

Then I heard no more words
Coming out from you
Then I heard no more words
Coming out from you
"Strange" is the song in our conversation
"Strange" is the song in our conversation

0

society

Posted by dayad on 21.14
It's a mystery to me
We have a greed
With which we have agreed
And you think you have to
Want more than you need
Until you have it all
You won't be free
Society
You're a crazy breed
I hope you're not lonely
Without me
When you want more than you have
You think you need
And when you think more than you want
Your thoughts begin to bleed
I think I need to find a bigger place
Cause when you have
More than you think
You need more space
Society
You're crazy breed
Hope you're not lonely
Without me
There's those thinking more less
Less is more
But if less is more
How you keepin score?
Means for every point you make
Your level drops
Kinda like you're startin from the top
And you can't do that
Society
Have mercy on me
I hope you're not angry
If I disagree
Society
Crazy indeed
Hope you're not lonely
Without me

0

menghibur diri

Posted by dayad on 10.36
menghibur diri dengan ngomong where there sorrow, there ease..
hahahaha.....lumayan

0

Pengenalan Suara Berbasis Neural Network dengan Algoritma Momentum Back Propagation

Posted by dayad on 20.50
Pendahuluan
Secara umum, kemungkinan akses ilegal pada suatu ruangan bertambah dengan bertambahnya jumlah pemilik akses. Ini karena media identitas -- seperti bar code, nomor PIN, kartu magnetik, dan kunci -- dapat diduplikasi. Dalam kasus ini, identitas yang melekat pada tubuh menjadi penting. Identitas yang melekat pada tubuh meliputi struktur DNA, retina mata, sidik jari, spektrum suara, dan lain-lain [1]. Dalam penelitian ini Kami menggunakan spektrum suara sebagai identitas untuk akses ruangan.
Alasan pemilihan spektrum suara sebagai identitas ialah berdasarkan fakta bahwa kita masih dapat membedakan suara dua orang berdasarkan suaranya walaupun mereka mengucapkan huruf vokal yang sama. Resonansi vokal dihasilkan oleh artikulator dalam mulut untuk membentuk suara vokal yang berbeda [2]. Karena artikulator hampir unik untuk setiap orang, maka kita bisa membedakan orang dari suaranya. Dalam penelitian ini kami memilih hanya menggunakan vokal "a" dengan alasan penyederhanaan masalah. Kami telah mempelajari bahwa spektrum suara untuk vokal "a" ialah berbeda untuk setiap orang.
Jaringan syaraf tiruan dengan algoritma pembelajaran Momentum Back Propagation digunakan sebagai voice recognizer. Jaringan syaraf tiruan dengan multi-layer perceptron digunakan karena cocok untuk pengenalan pola non-linear [3]. Input bagi neural netwok ialah spektrum vokal "a". Noise dari lingkungan telah direduksi menggunakan metode yang kami sebut self multiplication.

Momentum Back Propagation
Back propagation atau error back-propagation ialah teknik pembelajaran terawasi (superviced learning) yang digunakan untuk melatih jaringan syaraf tiruan. pertama kali dideskripsikan oleh Paul Werbos pada 1974, dan dikembangkan lebih lanjut oleh D.E. Rumelhart, G.E. Hilton, dan R.J. Williams pada 1986 [4]. Algoritma ini berdasar pada aturan delta umum (generalized delta rule):

dengan wij ialah bobot neuron pada layer-i ke layer-j, ialah laju belajar, dan ialah output SSE (jumlah kuadrat error).
Biasanya jaringan syaraf tiruan belajar lebih cepat dengan kenaikan laju belajar, tetapi dapat menghasilkan osilasi bahkan kegagalan belajar. Untuk mereduksi osilasi pada nilai laju belajar yang tinggi, faktor momentum [5] digunakan untuk menjaga kekonstanan kecenderungan proses belajar. Sehingga aturan delta umum untuk perbarahruan bobot menjadi,

dengan ialah faktor momentum.

Teknik Mereduksi Noise
Kami mengembangkan teknik mereduksi noise berdasarkan asumsi bahwa spektrum noise muncul secara acak. Dua sinyal yang mirip (dari sumber yang sama) yang direkam pada saat yang berbeda, tiap sinyal ditransformasi ke domain frekuensi menggunakan Fast Fourier ransform (FFT). Kedua spektrum hasil kemudian dikalikan komponen-per-komponen sehingga komponen spektrum yang muncul pada satu spektrum tapi tidak pada spektrum yang lain -- yaitu noise -- akan direduksi secarad dramatis. Metode ini kami sebut perkalian sendiri (self multiplication). Gambar 3.1 mengilustrasikan metode self multiplication dalam mereduksi spektrum noise.


Perhatikan bahwa metode ini menjaga komponen yang muncul pada kedua spektrum. Sehingga jika komponen noise muncul pada kedua spektrum, metode ini tidak bisa mereduksi noise tersebut.
[perhatikan bahwa noise yang kami reduksi ialah noise multi frekuensi, bukan noise single frekuensi]

Implementasi System
[bagian ini hanya diambil sebagaian saja, on request ]
Hardware untuk pengenalan suara meliputi unit komputer (dengan soundcard), microphone, dan sirkuit pengontrol motor berbasis mikrokontroller AT89S2051 yang terhubung ke PC melalui serial link.
Data base sample suara menggunakan suara dari 4 orang.
Berikut tampilan window di PC ketika berhasil mengenali suara.


Jaringan syaraf tiruan kami menggunakan 12 spektrum suara dari 4 orang sebagai referensi belajar. Training untuk 100 epoch dengan laju belajar 0.75, faktor momentum 0.75 dan kemiringan fungsi logistik 0.01 menghasilkan error global 0.065.

Ringkasan
Spektrum noise dapat dieliminasi menggunakan metode self miltiplication dengan beberapa keterbatasan. Metode ini membagi dua sinyal input (dalam domain waktu), mentransformasi tiap sinyal ke domain frekuensi, dan mengalikan kedua sinyal konponen-per-komponen.
Dalam online test (real case test), jaringan syaraf tiruan kami hanya berhasil mengenali satu dari empat orang secara baik (kepercayaan > 80%) berdasarkan spektrum suara, sisanya hanya dikenali dengan persentasi kepercayaan < 80%.

Referensi
[1]

Discovery Channel. 2057: The City. (a documentary movie)


[2]

Forming the Vowel Sounds. Available at: http://hyperphysics.phy-astr.gsu.edu/hbase/music/vowel.html


[3]

Anil K. Jain (Michigan State Univ.), Jianchang Mao and K. M. Mohiuddin (IBM Almaden Research Center). Artificial Neural Networks: A Tutorial. Pp. 39, March 1996.


[4]

Backpropagation. Available at: http://en.wikipedia.org/wiki/Backpropagation. Last modified: 24 June 2007.


[5]

Ben Krรถse (Fac. of Math. And Comp. Sci., Univ. of Amsterdam), Patrick van der Smagt (Inst. of Robotics and Sys. Dynamics, German Aerospace Research Establish). An Introduction to Neural Networks, 8th Ed. Pp. 37, November 1996.

sumber : http://www.forumsains.com/index.php?page=pengenalan-suara-berbasis-neural-network-dengan-algoritma-momentum-back-propagation


0

hal-hal yang harus di perbaiki

Posted by dayad on 07.52
ada beberapa point penting yang harus diperbaiki pada TA gw!! inget!!!
  1. cari teori yang pas untuk inventori
  2. customize teori tersebut buat di cocokkan dengan proses bisnis perusahaan
  3. benerin pada SWOT kata "dirasakan" ambigu dong
  4. benerin class diagram sama sequence diagram pada laporan dan prediksi
  5. ada kata2 yang salah
yah ini lah yg harus diperbaikin yad!! inget!!!

Copyright © 2009 dayad-t All rights reserved. Theme by Laptop Geek. | Bloggerized by FalconHive.