Computer Science - Science topic

Artificial Intelligence has been around for a long time – the Greek myths contain stories of mechanical men designed to mimic our own behavior. Very early European computers were conceived as “logical machines” and by reproducing capabilities such as basic arithmetic and memory, engineers saw their job, fundamentally, as attempting to create mechanical brains.

As technology, and, importantly, our understanding of how our minds work, has progressed, our concept of what constitutes AI has changed. Rather than increasingly complex calculations, work in the field of AI concentrated on mimicking human decision making processes and carrying out tasks in ever more human ways.

Artificial Intelligences – devices designed to act intelligently – are often classified into one of two fundamental groups – applied or general. Applied AI is far more common – systems designed to intelligently trade stocks and shares, or manoeuvre an autonomous vehicle would fall into this category.

Neural Networks – Artificial Intelligence And Machine [+]

Generalized AIs – systems or devices which can in theory handle any task – are less common, but this is where some of the most exciting advancements are happening today. It is also the area that has led to the development of Machine Learning. Often referred to as a subset of AI, it’s really more accurate to think of it as the current state-of-the-art.

The Rise of Machine Learning

Two important breakthroughs led to the emergence of Machine Learning as the vehicle which is driving AI development forward with the speed it currently has.

One of these was the realization – credited to Arthur Samuel in 1959 – that rather than teaching computers everything they need to know about the world and how to carry out tasks, it might be possible to teach them to learn for themselves.

The second, more recently, was the emergence of the internet, and the huge increase in the amount of digital information being generated, stored, and made available for analysis.

Once these innovations were in place, engineers realized that rather than teaching computers and machines how to do everything, it would be far more efficient to code them to think like human beings, and then plug them into the internet to give them access to all of the information in the world.

Neural Networks

The development of neural networks has been key to teaching computers to think and understand the world in the way we do, while retaining the innate advantages they hold over us such as speed, accuracy and lack of bias.

A Neural Network is a computer system designed to work by classifying information in the same way a human brain does. It can be taught to recognize, for example, images, and classify them according to elements they contain.

Essentially it works on a system of probability – based on data fed to it, it is able to make statements, decisions or predictions with a degree of certainty. The addition of a feedback loop enables “learning” – by sensing or being told whether its decisions are right or wrong, it modifies the approach it takes in the future.

Machine Learning applications can read text and work out whether the person who wrote it is making a complaint or offering congratulations. They can also listen to a piece of music, decide whether it is likely to make someone happy or sad, and find other pieces of music to match the mood. In some cases, they can even compose their own music expressing the same themes, or which they know is likely to be appreciated by the admirers of the original piece.

These are all possibilities offered by systems based around ML and neural networks. Thanks in no small part to science fiction, the idea has also emerged that we should be able to communicate and interact with electronic devices and digital information, as naturally as we would with another human being. To this end, another field of AI – Natural Language Processing (NLP) – has become a source of hugely exciting innovation in recent years, and one which is heavily reliant on ML.

NLP applications attempt to understand natural human communication, either written or spoken, and communicate in return with us using similar, natural language. ML is used here to help machines understand the vast nuances in human language, and to learn to respond in a way that a particular audience is likely to comprehend.

A Case Of Branding?

Artificial Intelligence – and in particular today ML certainly has a lot to offer. With its promise of automating mundane tasks as well as offering creative insight, industries in every sector from banking to healthcare and manufacturing are reaping the benefits. So, it’s important to bear in mind that AI and ML are something else … they are products which are being sold – consistently, and lucratively.

Machine Learning has certainly been seized as an opportunity by marketers. After AI has been around for so long, it’s possible that it started to be seen as something that’s in some way “old hat” even before its potential has ever truly been achieved. There have been a few false starts along the road to the “AI revolution”, and the term Machine Learning certainly gives marketers something new, shiny and, importantly, firmly grounded in the here-and-now, to offer.

The fact that we will eventually develop human-like AI has often been treated as something of an inevitability by technologists. Certainly, today we are closer than ever and we are moving towards that goal with increasing speed. Much of the exciting progress that we have seen in recent years is thanks to the fundamental changes in how we envisage AI working, which have been brought about by ML. I hope this piece has helped a few people understand the distinction between AI and ML. In my next piece on this subject I go deeper – literally – as I explain the theories behind another trending buzzword – Deep Learning.

Bernard Marr is a best-selling author & keynote speaker on business, technology and bigdata. His new book is Data Strategy. Toread hisfuture posts simply join his network here.

Thank you all, and my apologies for not thanking you guys earlier! To be frank I haven't logged in to Researchgate for a while and forgot I even asked this question...

But now that my thesis is done, I just might look into this a little bit further. The problem is still an interesting one and these resources have been really helpful.

Hi,

There is a site that provides the ranking of conferences by three sources, Australian, Brazil and Microsoft Academic.Here is the address

I tried to rank LNCS and it was ranked mostly on top of other conferences.

I follow answers

Mr. Nathan has rightly mentioned that work (energy) can be turned into an equivalent amount of information. Information can be turned into an equivalent amount of work (energy). Since high energy efficiency is required in a sensor network, it is desirable to disseminate query messages with small traffic overhead and to collect sensing data with low energy consumption.

Thank you Saif. With current dynamic research, innovation and technological advancement of Computing Technologies side by side with Bio-Technologies and Bio-Computing, is it critical to adopt and update the Computer Science and Engineering Curriculum to match the industry expectations from graduates as future potential employees.

Hi Alireza,

actually there are many ways to determine the similarity of signals (or the distance between them). Here some cues:

Using cross correlation (or correlation coefficient as normalized measure) surely is a widely applied method. As alternative correlation approach you could consider rank based correlation (Spearman's rho) (cross correlation e.g. assumes a linear dependency which might not be true).

Similarity of signals also can be accessed in the frequency domain, you could have a look for "coherence" to find more information.

Even by using measures of information theory similarity can be qunatified. Have e.g. a look to "entropy".

Apart from that, there are many distance metrics which can be applied similarly (the smaller the distance the bigger the similarity). The most often used metric is the euclidean distance (or a normalized version), but you'll find other metrics when searching for "distance measures".

Now, what's best suited strongly depends on your signals and what you regard to be similar (which can be problem specific). Hope this helps.

Greetings, Sebastian

TCP is a stream oriented protocol and UDP is a message-oriented protocol.

TCP receives the stream of bytes from application layer protocols and divide it in to segments and pass it to IP. But UDP receives already divided or grouped bytes of data from application protocols and add UDP headers which will become datagram and send it to IP, then application layers has the burden of dividing the streams of data in to messages when they run on top of UDP .

For signal processing you typically do:

Signal -> DWT -> Processing -> inverse DWT -> Output Signal

Where the processing step is doing whatever you would like with the DWT output. In the image you posted, you only show the Signal->DWT part, which gives you the DWT coefficients.

Referring to your image, the "A" (e.g. cA1) and "D" are the approximate and detailed coefficients from the transform. A simplified explanation would be, that "A" are the low-pass outputs, and the "D"s the high-pass outputs, but down-sampled by a factor of 2 at each stage (see wiki article).

What do you want to do with the DWT of the signal? Denoising?

I highly recommend you have a look at http://zone.ni.com/reference/en-XX/help/371419D-01/lvasptconcepts/wa_dwt/ where the whole signal processing with DWT is outlined. The image you show, is "only" for computing the decomposition coefficients, without processing the signal and reconstruction.

Because at every stage, a sub-sampling by a factor of 2 is performed, you need to put the outputs of the different stages together after up-sampling again for the inverse DWT and get the filtered version of your original signal in the time domain.

Let's look at a denoising example. To get rid of some noise in your signal, you may want to filter out some high-frequency components of your signal. At the end of the image you show, we have the DWT coefficients. At this point, we could reconstruct the original signal with these coefficients, no processing is done. But now, you can e.g. throw away some high frequency coefficients to reduce the noise. So now you are left with a reduced number of coefficients / subbands. If you want to get the processed signal again in the time domain, you need to apply an inverse DWT:

The link I provided above shows this nicely.

So long story short, in the image you posted, you only show the Signal->DWT part, giving you the DWT coefficients, and not yet a filtered version of your input. From these coefficients, you can reconstruct the original signal using the inverse DWT. If you want to process, do something with the coefficients before applying the inverse DWT to get the signal in the time-domain.

Here's some example code for C++ if you want to play around a bit:

I hope that clarifies it a bit.

Please let me know in detail, your Research Work.

I am a PhD Guide at The Maharaja Sayajirao University of Baroda.

I thank you all for your answers.

Very interesting academic debates and insightful arguments! Great work done all commenters on the subject. Both intellectual and practical knowledge are all important in establishing and defending truth. They both aid in the discovery and application of knowledge for the development of societies.

Dickson Adom

Dear Esmat,

Check The link below (Asian Journal of Information Technology): USD 250 for fees, fast publication, and Q4.

Dear Rakesh,

this question looks similar to the one at this link: https://www.researchgate.net/post/Can_someone_please_suggest_some_good_Scopus_indexed_fast_journals_where_I_can_publish_my_research_paper_in_quick_time. Please, take a look at it, since there are good hints and resources listed there.

Hope it helps.

Best,

Luca

Virtual reality (VR) is a new technique. It is still in experimental stages in Radiology and imaging. It is mainly used to see and analyse the DICOM images on a tablet or mobile phone. VR can be as useful in Pediatric Radiology as in other sub specialties of Radiology. This technique may be utilized to read the fetal/infant brain MRI images on mobile phones and tablets.

In computer science, this is known as inter-process communication.  More specifically, when a process spawns a sub-process, it is referred to as a child process.   A parent process and a child process can communicate using an Unnamed Pipe.  Independent processes (on the same machine) can communicate using Named Pipes.

NS2 is no longer used and active development stopped 7 years ago. So its best you use NS3 or NetSim or OPNET. And like you say you need to have lots of time and be very good at programming (in multiple languages) for Ns3.

I am not from any field of Engineering but I feel Human - Computer Interaction is the current and upcoming field and can be taken up for research. There can me may aspects of this field - you can take the Economic, Social, Administrative and other such aspects to make research valuable for the society/today's world.

FFT can be used....

Here you can find brain MRI dataset (some include white matter lesions). Hope it helps:

Dear Asiimwe,

I suggest you to see attached lifes and publication in topics. 

Best regards

You need to make sure your input data is normalised as training may not occur if you don't normalise input data.

As you are doing a classification problem  you will need more than one output node. If you only have one output node and you assign colour A if the value on the output node is  0 to 0.25, colour B if the value is 0.25 to 0.5, and colour C if the value is 0.5 to 0.75 and colour D if the value is 0.75 to 1, then you will be solving a regression problem, not a classification problem.

So you need two output nodes and output (0,0)=colour A, output (0,1)=colour B, output  (1,0)=colour C and output (1,1)=colour D.

Or you can have four output nodes as output then (1,0,0,0)= colour A, (0,1,0,0)= colour B. etc

Now the actual values you choose at the nodes depends on the output function you have chosen. The function may range from 0 to 1, in which case you use the above values or your output function may range from  -1 to 1, in which case replace the 0 by -1 in the above and keep the 1 as it is. Or your output function may have a range -0.5 to 0.5, in which case the zeros are replaced by -0.5 and the ones replaced by 0.5 in the above brackets.

Dear all,

Thank you for the interested you expressed in our project, and for all your comments.

The project is in its seminal stage, and this is why I have not updated more details just yet. I realise now, it was perhaps a mistake to post just a title in the first place, given the state of our research is in, and I apologise for that.

Regarding the request for updates I have received lately, allow me to say that I'd like to reserve the right to post more details when I am comfortable that the results I have are robust enough to be shown. I hope you will understand.

What I can say however is that the project will present a number of novel approaches, techniques and theories to face new problems brought about by the pervasive use of robotics and autonomous systems in different domains of human society. This project aims also, and in turn, to provide Philosophy with a laboratory for testing the validity of its theories at a larger-scale, as opposed to the current philosophical tradition of small-scale thought experiments. 

The project definitely does not:

"attempt to nudge philosophers into reflecting on the significance and underlying intent of machine-oriented systems such as decision-making robots"

On the contrary, we use philosophical theories to develop novel computational techniques, programs that work, like physics or mathematics are used in computing science with the same goal. Therefore, no Philosophy of <x> will be proposed or discussed, the level of abstraction (Floridi, 2008) is much lower than that.

The project is definitely not about:

"HOL (Higher Level Logic) in a proof-theoretic approach to machine behaviour"

We take a precise position about the use of logic as approach to a certain subset of AI problems, in fact we depart from that tradition showing that probabilistic methods are the most suitable solution, and do work very well in Philosophy too.

Regarding what Muna Alsallal said:

"I think machine learning ML can be assumed as a trend that can change the ordinary aspects of the world"

I am not sure what is the role of this assumption with regards to the original question (i.e. What is this project exactly? I do Philosophy and Computer Science myself so I'm curious to see a hybrid like this).

I do honestly apologies for the vagueness of this answer. I am truly delighted that you are interested (or curious) to know more about our project, and I please you all to bear with me a bit longer. I'll post more details very soon. In the meanwhile, I'd be happy to answer any question you might have, and keep the discussion alive at any level you prefer.

With kind regards,

Francesco Perrone

--

Floridi, Luciano. "The method of levels of abstraction." Minds and machines 18.3 (2008): 303-329.

Dear Kelvin,

The equation [x/a] + [y/b] = c, where  [  ] is the floor function, has the 

general  solution:  x = ma + r and y = (c-m)b + s,

where m is any arbitrary integer,  0 ≤ r < a and  0 ≤ s < b.

Now, you need x = y  > 0  to obtain positive  solutions x for  [x/a] + [x/b] = c ,

direct substitution shows m = ( bc + s - r )/(a+b).

Next, you test all possible values of s and r to give a list of all possible positive integer values of m. Otherwise the equation has no solution.

Example :  [x/2] + [x/3] = 5.

m = ( bc + s - r )/(a+b) = ( 15 + s - r )/(5)  where  0 ≤ r < 2 and  0 ≤ s <3.

All possible values of m are {3,14/5,13/5,16/5} , m= 3 is the only positive integer solution.

Therefore x = ma +r  provides  6 and 7  are all possible solutions for   [x/2] + [x/3] = 5.

Best wishes 

Dear Dr. Peter,

            The size is one of the useful things about the elephant. But they also have other uses. Like dogs, they can be bred to be ferocious, loyal and intelligent. These characteristics are also very useful in battle. Likewise, horses also have many desirable characteristics. Pigs or baffles cannot tolerate the noise or pain. They would turn back and trample our own soldiers. Likewise, components are versatile and well proven to have many subtle uses.

          You have no imagination and have no open mind. You are incapable of overcoming prejudice and preconceived notions. You don’t want to try new possibilities or learn new things. Even stanch religious conservatives would explore new ways.

           You said: Your idea is closest to being true for web page designers. They do select from a palate of options and drop them in to their design, stretching or cutting to fit.

            I said many times to you: The most complex software (in my view) that is least friendly to be componentized for CBD are compilers. I told you many times, even it is possible to implement large percent of code for compiler in pluggable components.

            Almost every text book on structure or records use Employee as an example. Is structures or records are only capable of creating records for Employee? Many of the GUI components (e.g. Pie-chart, line-chart or bar-chart) are demoed using financial data or stock movements. Are the GUI components such as line chart only useful for presenting intraday stock movements?

            Stop insulting me and trying to put me down. If you can’t accept the reality, it is your problem. The reality is that: Neither complexity nor uniqueness of a (physical or software) product can’t prevent its designers to partition the product in to self-contained parts that are conducive to be assembled.

              Almost every physical product is partitioned into self-contained parts that are conducive to be assembled. Nearly 100% of the features and functionality can be implemented in self-contained parts that are conducive to be assembled.

               Not even a single software product can be partitioned into self-contained parts that are conducive to be assembled. No attempt is made to partition even 10% of the features and functionality in self-contained parts that are conducive to be assembled. No one can help you, if you don’t want to accept this reality.

               It is a common sense: one would use any other animal, if it is loyal, ferocious, large and can be domesticated. We want to have animal that is having more useful characteristics for the task.  Did you see the movie Avatar? They have animals that are having better characteristics then houses that can fly. If you can invent a new kind of parts that are better than the components in modularization and/or other uses, then you can name them Wudgets or whatever you like. They can provide different but more useful then the components.

Bye,

Raju

Its been years and we have upgraded our services and site to offer Cloud Web hosting, Wordpress hosting, Managed Servers, Ecommerce Hosting , Domain registration and many more .. IF you are thinking of online Business dont forget to check out : https://www.Webhost.UK.Net

Trade-offs:

Speed v texture ["depth" can be illusory; detecting interactions and subtle dynamics allows understanding of data flavor but takes time]

Stability v Adaptability [are criteria clear and stable? is it important to detect emergent or unstable characteristics?]

GAs are faster; standard stats are more easily applied [and better understood]; recurrent hierarchical networks are slower and require more intense design but offer granularity and flexibility and can be audited [to a limited extent]

Analyse existing used matrices  for  different measure of component based software and try to find gap between measures and try to develop metrics which will fill those gaps between measures   

Dear Researcher/Dear Colleague,

This work is a part of wider project that deals with promoting of complex systems, self-organization and emergence. For me and my coworkers it will be great to hear what you find good and not so good on it and enable us to deliver more suitable projects in the future.

Good luck at your research,.

Jiri Kroc

PS Another interesting questions are involved in the project as is:

@Nazanin "A graph peoperty is monoton if every subgraph of a graph with property P, also has the property P"  is called "hereditary property".  

As for the original (part of a) sentence: if we join it to another domain of mathematics, it makes sense such as it is, despite imprecise wording: e.g. in a class of exponential functions with a basis a>1... 

Informatica..

People interested in an update on predatory journals, following the cessation of Beall's List updates and it's removal, might like to refer to an article in today's Times Higher Education (6 April 2017, pp. 42 & 43), reporting work by Larissa Shamseer and David Moher (link below but access may be restricted to subscribers).

Their study resulted in a list of 13 warning signs:

1. The scope of interest includes non-biomedical subjects alongside biomedical topics, or multiple, wide-ranging and unrelated fields of study are combined

2. The website contains spelling and grammar errors

3, Images are distorted/fuzzy, intended to look like something they are not, or are unauthorised

4.The home page language targets authors rather than readers

5, The Index Copernicus Value is promoted on the website

6.  Description of the manuscript handling process is lacking

7.   Manuscripts are requested to be submitted via email

8.   Rapid publication is promised

9.   There is no retraction policy

10.   Information on whether and how journal content will be digitally preserved Is absent

11. The article processing/publication charge is very low (eg, <$150)

12. Journals claiming to be open access either retain copyright of published research or fail to mention copyright

13.  The contact email address is non-professional and non-journal affiliated (eg, @gmail.com or @yahoo.com)

Look at how the Eigen, Catch and the Boost C++ libraries started as pet project of one or a few developers and are now widely used even in big companies. There are lots of examples contradicting the "millions of dollars needed to start" model, maybe you should look at how they did it in more details. 

Of course, it will happen over years and Google won't use your stuff next month but in the meantime, your ideas will spread and your library improve with the real feedback you will get from real users. Had you gone this path a decade ago, you might have been one of the library I was giving as an example!

Good luck,

Bruno

Hey Guys I need help,

Every time I tried to convert the nessus vulnerability scanning file to nessus.p file which is the input for mulval I am getting below error

++Error[XSB/Runtime/P]: [Existence (No procedure usermod : vulExists / 3 exists)] []

Forward Continuation...

... machine:xsb_backtrace/1 From C:\Program Files (x86)\XSB\syslib\machine.xwam

... loader:load_pred1/1 From C:\Program Files (x86)\XSB\syslib\loader.xwam

... loader:load_pred0/1 From C:\Program Files (x86)\XSB\syslib\loader.xwam

... loader:load_pred/1 From C:\Program Files (x86)\XSB\syslib\loader.xwam

... x_interp:_$call/1 From C:\Program Files (x86)\XSB\syslib\x_interp.xwam

... x_interp:call_query/1 From C:\Program Files (x86)\XSB\syslib\x_interp.xwam

... standard:call/1 From C:\Program Files (x86)\XSB\syslib\standard.xwam

... standard:catch/3 From C:\Program Files (x86)\XSB\syslib\standard.xwam

... x_interp:interpreter/0 From C:\Program Files (x86)\XSB\syslib\x_interp.xwam

... loader:ll_code_call/3 From C:\Program Files (x86)\XSB\syslib\loader.xwam

... standard:call/1 From C:\Program Files (x86)\XSB\syslib\standard.xwam

... standard:catch/3 From C:\Program Files (x86)\XSB\syslib\standard.xwam

Any idea how to solve this error??

@Steven: I found it, thanks. I have also had a look at Armstrong's website. There is quite a number of publications, did you have some specific paper in mind?

To a great extent it depends on the class.  Graph theory, linear algebra, generating functions, logic, set theory, and some basic number theory as well as basic probability theory will  help a lot. Especially if you want to go on for graduate school. If you are interested in crypto for instance some group theory and more advanced probability is needed. Machine learning is all about probability and statistics. Formal methods and programming languages require abstract algebra and lots of logics.

 Please see the database I uploaded here on RG:

I hope this helps.

Dear Friends,

         Let me provide clear and unambiguous contrast between the self-contained components for software and physical products (no contradicting can be as black and white as this contradiction): In the context of the physical products (every large physical product can be partitioned into many large disjoint sets of self-contained features and functionality): Almost each and every disjoint set of self-contained feature and functionality can be implemented as replaceable component, which can be disassembled and reassembled at any time in the future (e.g. redesign and test individually free from spaghetti design). That is, each component is assembled or plugged-in as illustrated in FIG-1.

          In the context of the software products (every large software product comprises many large disjoint sets of self-contained features and functionality): Almost no large disjoint set of self-contained features and functionality is implemented as replaceable component (e.g. as a module that can be assembled or plugged-in). But each and every large disjoint set of self-contained features and functionality can be implemented as RCC (Replaceable Component Class).

That is, the code base of each large self-contained component is broken into many code chickens and added to (or spread across) many non-exclusive files (i.e. mixed with code chunks for other self-contained components), where each non-exclusive file contains spaghetti code for multiple large self-contained components (as illustrated by FIG-2, FIG-3 & FIG-4 in the attached PDF).

         The root cause for this problem is: No one in the world even today knows what is the purpose and true essence of the CBD for physical products. The flawed “consensuses” reached 50 to 60 years ago diverted research efforts into a wrong path, as the 2300 years old flawed consensus “the Earth is static” diverted research efforts into wrong path that resulted in geocentric paradox (an altered perception of reality filled with epicycles and retrograde motions).

           My struggle to compel research community to investigate evidence, reality and facts to discover the nature and essence of the real-CBD (Component Based Design) for physical products, which can be summarized as: CBD is Implementing about 90% of the features and functionality in replaceable components, where replaceable component imply components that can be easily dis-assembled (or un-plugged to redesign or refine individually outside of the product free from spaghetti design/code) at any time for periodical feature or functionality maintenance or up-gradation releases and re-assembled (or re-assembled after testing it individually free from spaghetti code). Since over 90% of the design and code for features and functionality is implemented in such components (where each is free from spaghetti code), over 90% of the design and code of the product is free from the spaghetti code.

         Such real-CBSD addresses the infamous software crisis, because the root cause for the software crisis is the infamous spaghetti code. Design and development of no other product is infected with such spaghetti code, for example, particularly design and development of complex one-of-a-kind-products such as experimental spacecraft or fully-tested pre-product golden prototypes of next generation Jet-fighters. It is not hard to achieve such real-CBD: https://www.researchgate.net/publication/292378253_Brief_Introduction_to_COP_Component_Oriented_Programming

      Today no one even knows what the true essence of the CBD is, but every researcher insist that it is impossible to invent real-software-components for real-CBD. I have been struggling to provide counter evidence for many years such as hundreds of software applications built by assembling replaceable components (e.g. See real-software-components.com and pioneer-soft.com).

       Any research efforts for any discipline would be diverted into a wrong path, as soon as research efforts start relying of flawed “consensus” (by assuming then to be self-evident facts). The research efforts persist in the wrong path for long enough time (without realizing the error), the research efforts accumulates huge BoK (Body of Knowledge) filled with epicycles over the time and the discipline ends up is a paradox. Today, exiting CBSD paradox is supported many times bigger BoK than the BoK that existed 450 years ago in support of geocentric paradox.

Best Regards,

Raju Chiluvuri

If it is about cplex (see https://en.wikipedia.org/wiki/CPLEX ):

get the software bundle - including the libraries (binaries) for your compiler and OS and follow the documentation about installation/use.

Having installed the libraries correctly, you have already 'added' them to your compiler. Then - as Peter Breuer already stated - you have to include the appropriate header files to be able to call the functions contained in the libraries.

Really "Compiler 101".

if you mean multiclass classification using SVM, see 'fitecoc' in matlab:

Hello Aditya,

What are the Objects you want to recognise ? Are they same object? and more in Number?

Example : 1st Case : You want to Detect Multiple objects of same class .

You want to detect Multiple Oranges in a Picture.

2nd Case : Multi Class Object detection

You want to detect Oranges, Apples, Banana's . 

If you are looking for the first case, it should not be that hard to implement, Apply the object detector multiple times on the images, instead returning when you find the first object.

If its the second case, try building feature detectors for every class and apply every detector onto the image to find the different objects.

In general there are many approaches to recognise the object. Simple features like Color, Shape, or more complex features using HaarWavelets.

Start with one object and move to multiple object recognition

Hi Veera, 

I believe that your dataset has no meaning in the current structure. My suggestion is to change each column into a raw (except the last column) considering the associated "percentage" attribute with each subject (if the "prcentage" attribute belongs to the subject). Afer that, if your'e using WEKA, then evaluate your learning process based on "Percentage split" method becaues cross-valiation will not work.

NB: Increase the amount of your dataset, as you won't get any useful model from it to be used in the future. 

HTH.

Samer

Thank you sir i will refer the links

Dear Mr. Krish,

            The biggest problem in case of scientific crisis is exposing the error (i.e. unproven myth “the Earth is static”) at the root of the geocentric paradox. The unproven axiom (i.e. “the Earth is static”) diverted scientific progress into a wrong path, which altered perception of reality so much so, the Truth was perceived to Heresy or repugnant.

            Although, there is enough evidence exists by 17th century, research community prosecuted Galileo for saying that the Earth is not static. Discovering the Truth was not hard, once all the evidence is investigated with open mind. The truth could have been discovered in normal course of investigation, even if the truth were to be “Saturn is at the center” – By just exploring each possibility by plotting the planarity positions by putting each planet.

            If you are me wanted to discover the Truth with open mind, what would we do?  If I think, Saturn is at the center, I try to collect as much evidence as possible and try to plot planetary paths and try to predict future locations to validate my prediction.  

            Remarkable similar kind of problem repeated in the computer science. Software researchers defined 50 years ago that CBD is building software by assembling reusable (or COTS – Commercially off the Shelf) components, as hardware engineers build computers by assembling COTS from 3td party component vendors. Also few researchers even called them Software-ICs.  – This has no basic in reality/fact, but software researches have been perusing this wrong path for past 50 years.

            Since they have been perusing a wrong path for over 50 years, it altered perception of reality so much so, the Truth was perceived to Heresy or repugnant. This error may have cost a trillion dollars during past 25 years and would cost trillions more, if the error is not exposed. The reality of CBD is explained at: https://www.researchgate.net/publication/284167768_What_is_true_essence_of_Component_Based_Design. I accomplished this goal many years ago and trying to expose the error. My CBSD is not correct, the Truth can be discovered by any one by investigating all the evidence.

            It was impossible to discover the Truth by prosecuting any one, who questioned the validity of flawed seed axioms (i.e. the Earth is static). I have been enduring insults, personal attacks and humiliation for requesting proof for flawed concepts such as “reusable software parts are components, and CBSD is using such fake components”. The Truth can be discovered within couple of weeks, if the can be exposed.

The biggest problem is admitting or acknowledging the error. The “scientific method” requires investigating all the evidence with open mind and going wherever facts lead us, without being influenced by prejudice and preconceived notions.

Quotes by Arthur Schopenhauer (Great 19th Century German Philosopher)

“All truth passes through three stages. First, it is ridiculed. Second, it is violently opposed. Third, it is accepted as being self-evident.”

“The discovery of truth is prevented more effectively, not by the false appearance things present and which mislead into error, not directly by weakness of the reasoning powers, but by preconceived opinion, by prejudice.”

            I didn’t think, they are true even in 21st century. But based on firsthand experience, they are true even in 21st century.  No software experts can support their opinion, except by saying every one else things reusable software parts are components. It is not science. Science needs objective proof.

            In an open scientific debate, one must back each fact by proof and evidence. I can back each of my invention by irrefutable proof and evidence. Also I can prove existing concepts are flawed by using irrefutable proof and evidence. I am requesting for an opportunity to demonstrate proof, but no one is willing to give me an opportunity. I put all the evidence openly in my web sites as well as tools for building real CBD applications.

                      Compared to problems such as “Einstein's constant velocity of light”, the problem we in software trying to solve is elementary school math.

Best Regards,

Raju

Hi Vishal

Any system that is developed will be no faster than the weakest link. So you need to search for the bottleneck, identify where it is at, and work out how to deal with it.

So, in your case, if the bottleneck is at the IoT device, there is likely nothing to be gained from making improvements in the connection between the gateway and the cloud. And it may be the case that there is no point in improving the connection between the IoT device and the gateway.

You need to start by analysing every component in your system in order to find out the maximum capacity it can consistently deal with. You then need to assess the capability of the connection between each component of the system and then next in the chain, to establish what the maximum throughput capacity is, and finally the throughput to the cloud.

Think of it like you would a large river system. High in the mountains, rain falls, or snow melts and the water drains into the ground. Eventually it becomes waterlogged and it starts to form a little tributary, many of which will flow into a stream, which grows bigger and bigger the further the system flows. If the ground is very dry, the water is instantly absorbed into the ground, and there is no flow downstream. Liken this to your IoT sensors. The barrier to information flow here is the dry ground which diverts the flow. Once you overcome this problem, when the ground becomes waterlogged, tributaries start to flow. Liken this to resolving the congenstion at the IoT devices. Many tributaries, covering a wide area eventually flow into streams. Liken this to the aggregation of data collection from many IoT devices. Let us say there is a dry lake bed in the way. Liken this to a congenstion between IoT device collection and the gateway. Until the lake is full, the water will not flow past it. Once the lake fills, water will flow over the edge into a waterfall. Liken this to the congestion between gateway and cloud. Once the water flows beyond this point, nothing can stop it until it reaches the sea. Liken this to outflow from the gateway to cloud connection being de-congested, where the cloud can be expanded to absorb all data flows.

Given the infinite scalability offered by cloud services, that end is unlikely to be at issue. Simply add more resources, whether compute, or storage, or network and you will handle whatever you need to throw at it. So identify the specific capacity of each of these elements in your cloud architecture to discover what your maximum flow rates are. Now, if we liken the waterfall to the internet, check your maximum flow rate through the internet. If this is less that the cloud you have specified, either change internet supplier to a faster package, or reduce the capacity of the cloud services you are running to match.

Now repeat this exersise all the way through to the individual IoT data sensors to identify what the various maximum flow rate capacities are, the total number of these feeding into your system, and check your sums to find the bottleneck.

If the bottleneck is at the IoT sensors, perhaps you either have sensors that don't have enough capacity, or you are sampling your data too frequently. IoT sensors are very cheap, and it may be that the software you are running on them is too bloated. You may be able to find much more compact software. Or you can replace them with better capacity sensors, or you can drop your data collection frequency.

If the bottleneck is between the sensors and the gateway, then you need to analyse traffic patterns. If the data comes in high and low spurts, you could do data collection, then re-transmit collected data in a steady stream to the gateway. Or if you have too many sensors, you need to consider whether you need additional gateway capacity, or could you use the data collector to spread the load.

As for the congestion between gateway and cloud, you can add aditional resources as needed to solve the problem.

The bottom line is you need to measure all the throughputs of all the components so that you are aware of what your spare capacity will be throughout the system architecture. Some components will be far more expensive to scale than others. So you need to do your sums to discover where the cheapest point would be to expand capacity. The IoT has an infinite capacity to create data, so finding the most cost effective route to acquiring and collecting that data is essential. Equally, you do not want to be paying for more cloud services than you absolutely need, especially where the systems will run 24/7 365 days a year. The costs can really stack up.

I hope this gets you started.

Regards

Bob

netc = closeloop(net);

[xc,xic,aic,tc] = preparets(netc,X,{},T);

yc = netc(xc,xic,aic);

tcs = size(tc,2);

closedLoopPerformance = perform(net,tc,yc)

where 'X' and 'T' are input and target vectors

'net' is narx network that is already created by you

There is no such best feature extraction technique(PCA, KPCA, Markov model) for malware detection....What type of features of malwares you want to detect or classify that need to be considered.....You may follow the follwing links.....

file:///C:/Users/Sanjay/Downloads/ELK-1601-189_manuscript_1.pdf

I think you can try to use any of the methods using for generate an initial population:

- Random (Soumitra K Mallick)

- Quasi-Random

- Secuential

- Parallel (Soheila Ghambari)

- Heuristic (Dênio Júnior)

Please go through the web site:

You can see this website for learning lot of concepts related to computer science.

Hi Neha..see following can be useful.

1) Parallelize Algorithms for XXX. Now you may complete the sentence by replacing XXX for a specific architecture, parallel processing platforms, distributed processing platforms etc.

Dear Amin Bakhtiari,

In the case of single objective GA, please refer to the CEC2005/2013 which contain problem definitions and evaluation criteria for the single objective real-parameter optimization problems. Otherwise, refer to the CEC2009 multiobjective optimization test instances.

We have both tried in various ways to guide you to RANKING the data ..... you have n raw scores from say large negative to large positive numbers (can be decimal numbers). The rank of each score is its place in an ordering from say LARGEST (1) to SMALLEST (n)  (or the other way if your interpretation suits). The point is that there is almost nothing wrong you can do -- have the courage of doing your own analysis rather than trying to copy another paper? especially when the steps seem pretty obvious.... note that the published example you show clearly says that they have used ranks, and the resultant are numbers that are all different by a few integers....

I think it would be interesting if you were trying to interpret the magnitude of the gaps between the scores for two observations on two different scales there might be some work to do to normalize the values --- clearly a very large range in the underlying data could appear to give a very large difference on one scale that might not be mimicked on the other; if you are interested in the relationship or coherence between the different metrics, you could use the raw data in a correlation analysis (excel will do this -- see the data analysis add-in)

Hope this helps.

Hi Yasin

There will be a number of issues that will need to be addressed here. Ideally, you would want to do it in as cost effective a manner as possible. First, the obvious option would be to see if you can find a nearby power source, either to drive the hardware, or to top up the charge in the battery. This is often not practical, such as in IoT at remote locations, or remote SCADA system elements. The next logical choice would then be to see if you can install a solar powered feed to continuously recharge the battery. A water powered generator would also fit the bill here. Assuming these options are impractical, then increasing battery capacity would be the next logical move, either by research to provide greater capacity (very expensive) or by installing a bigger battery. If these turn out not to be practical, then you need to look very carefully at your hardware to see whether it is matched to the demands being placed on it, and whether the software running on it is over bloated, or signals are being passed through it with too high a frequency. Matching CPU capability to load would be a useful start. Minimising the operating system and software code running on the hardware would also be a smart move. Most operating systems and software running thereon are heavily over-bloated. If you are running something with 150MB of code, but you are only using a very small portion of this, then by running exactly what you need, there will be a much smaller draw on your available battery capacity. Similarly, cutting down the frequency of signals being passed though the hardware would also reduce the load. You could create a temporary data store on a high speed SD card, then update once a day instead of doing live updates, subject to your needs. Usually, you can cut down on frequency without too much impact on the overall system functionality. These last three suggestions you can carry out now, with minimal cost outlay. All the previous suggestions will involve some element of capital spend and/or research which will take some time to gain any positive benefits.

Hope that helps.

Regards

Bob

It could also be that the plates were warm when they went into the chiller causing condensation in the bag and plates.  This could cause the agar to get wet and dilute it out the agar if it had not solidified well when first put in the chiller.  I have had this happen before and only put plates in the refrigerator after they have completely cooled most times waiting over night to bag them and chill them down. 

Have you tried contacting them directly? It might work faster.

Hi Yasin,

Good to know that you are willing to contribute a book. First of all know ur target audience, whom u r writing d book for, design the table of content or index for ur reference. So that u r clear with what chapters covers what content precisely...

First of all know ur target audience, whom u r writing d book for, design the table of content or index for ur reference. So that u r clear with what chapters covers what content precisely...

design the table of content or index for ur reference. So that u r clear with what chapters covers what content precisely...

Read the existing literature and quote them well. 

Write simple english; understood by all. As if u r teaching someone...it should be in that such simple format,  explain it with may be some cases, examples, etc. for better understanding.

All the best. Will be looking forward for your book

Mukta

I think what Knuth means is that you should not start optimizing because of a "gut feeling" that some algorithms in the program may run faster. You should first finish the program and then measure the performance. If the performance is sufficient, stop; else identify the problematic parts and optimize only as much as necessary.

Faster code is less readable, and a non-critical perceived problem probably solves itself in one or two processor generations (or your optimization becomes obsolete), so keeping it simple is a very good advice. As Peter implies, the actual most critical parts are often in the most unusual and unexpected places.

So the message is to base the optimization decision on facts (measurements) and not on guessing (programmer intuition).

A Prefetch Algorithm

Dear, check this link:

First you need to find proxies for IT and SI. These will be representative of each IT and SI. Your choice could depend on availability of data as well as previous research among other factors.

Once you have your variables, you can conduct all kinds of association analyses including but not limited to pair-wise correlation, scatterplots, regression, etc.

You might be interested in this paper:

thank you Mr.Giorgio Roff for your replay 

There are not very easy for children.

Hojjat Allah Bazoobandi answer is very correct

Dear Manish,

In the following related papers you can find many relevant feature sets:

(1)

Sedaaghi, M. H. (2009). A comparative study of gender and age classification in speech signals. Iranian Journal of Electrical and Electronic Engineering, 5(1), 1-12.‏

(2)

Lingenfelser, F., Wagner, J., Vogt, T., Kim, J., & André, E. (2010). Age and gender classification from speech using decision level fusion and ensemble based techniques. In INTERSPEECH (Vol. 10, pp. 2798-2801).‏

(3)