MLT Unit 4 Part 1 Artificial Neural Network and Deep Learning

Que4.1. Describe Artificial Neural Network( ANN) with different layers. Answer Artificial Neural Network ReferQ.1.13, runner 1 – 14L, Unit- 1. A neural network contains the following three layers Input subcaste The exertion of the input units represents the raw information that can feed into the network. retired subcaste retired subcaste is used to determine the exertion of each retired unit. ii. The conditioning of the input units and the weights depend on the connections between the input and the hidden units. iii. There may be one or further retired layers. Affair subcaste The geste
of the affair units depends on the exertion of the hidden units and the weights between the retired and affair units. Que4.2. What are the advantages and disadvantage of Artificial Neural Network? Answer Advantages of Artificial Neural Networks( ANN) 1. Problems in ANN are represented by trait- value dyads. 2. ANNs are used for problems having the target function, affair may be separate- valued, real- valued, or a vector of several real or separate- valued attributes. 3. ANNs learning styles are relatively robust to noise in the training data. The training exemplifications may contain crimes, which don’t affect the final affair. 4. It’s used where the fast evaluation of the learned target function needed. 5. ANNs can bear long training times depending on factors similar as the number of weights in the network, the number of training exemplifications considered, and the settings of colorful literacy algorithm parameters. Disadvantages of Artificial Neural Networks( ANN) 1. tackle dependence Artificial neural networks bear processors with resemblant processing power, by their structure. b. For this reason, the consummation of the outfit is dependent. 2. Unexplained functioning of the network a. This is the most important problem of ANN. b. When ANN gives a delving result, it doesn’t give a indication as to why and how. c. This reduces trust in the network. 3. Assurance of proper network structure There’s no specific rule for determining the structure of artificial neural networks. b. The applicable network structure is achieved through experience and trial and error. 4. The difficulty of showing the problem to the network ANNs can work with numerical information. Problems have to be restated into numerical values before being introduced to ANN. c. The display medium to be determined will directly impact the performance of the network. d. This is dependent on the stoner’s capability. 5. The duration of the network is unknown a. The network is reduced to a certain value of the error on the sample means that the training has been completed. b. This value doesn’t give us optimum results. Que4.3. What are the characteristics of Artificial Neural Network? Answer Characteristics of Artificial Neural Network are 1. It’s neurally enforced fine model. 2. It contains large number of connected processing rudiments called neurons to do all the operations. 3. Information stored in the neurons is principally the weighted relation of neurons. 4. The input signals arrive at the processing rudiments through connections and connecting weights. 5. It has the capability to learn, recall and generalize from the given data by suitable assignment and adaptation of weights. 6. The collaborative geste
of the neurons describes its computational power, and no single neuron carries specific information. Que4.4. Explain the operation areas of artificial neural network. Answer operation areas of artificial neural network are 1. Speech recognition Speech occupies a prominent part in mortal- mortal commerce. thus, it’s natural for people to anticipate speech interfaces with computers. c. In the present period, for communication with machines, humans still need sophisticated languages which are delicate to learn and use. To ease this communication hedge, a simple result could be communication in a spoken language that’s possible for the machine to understand. Hence, ANN is playing a major part in speech recognition. 2. Character recognition It’s a problem which falls under the general area of Pattern Recognition. b. numerous neural networks have been developed for automatic recognition of handwritten characters, either letters or integers. 3. hand verification operation Autographs are useful ways to authorize and authenticate a person in legal deals. hand verification fashion is anon-vision grounded fashion. c. For this operation, the first approach is to prize the point or rather the geometrical point set representing the hand. With these point sets, we’ve to train the neural networks using an effective neural network algorithm. e. This trained neural network will classify the hand as being genuine or forged under the verification stage. 4. mortal face recognition It’s one of the biometric styles to identify the given face. It’s a typical task because of the characterization of “non-face ” images. still, if a neural network is well trained, also it can be divided into two classes videlicet images having faces and images that do not have faces. Que4.5. Explain different types of neuron connection with armature. Answer Different types of neuron connection are 1. Single- subcaste feed forward network a. In this type of network, we’ve only two layers i.e., input subcaste and affair subcaste but input subcaste doesn’t count because no calculation is performed in this subcaste. Affair subcaste is formed when different weights are applied on input bumps and the accretive effect per knot is taken. c. After this the neurons inclusively give the affair subcaste to cipher the affair signals. 2. Multilayer feed forward network a. This subcaste has hidden subcaste which is internal to the network and has no direct contact with the external subcaste. Actuality of one or further retired layers enables the network to be computationally stronger. c. There are no feedback connections in which labors of the model are fed back into itself. 3. Single knot with its own feedback a. When labors can be directed back as inputs to the same subcaste or antedating subcaste bumps, also it results in feedback networks. intermittent networks are feedback networks with unrestricted circle. 4.5.1 shows a single intermittent network having single neuron with feedback to itself. 4. Single- subcaste intermittent network a. This network is single subcaste network with feedback connection in which processing element’s affair can be directed back to itself or to other processing element or both. intermittent neural network is a class of artificial neural network where connections between bumps form a directed graph along a sequence. c. This allows it to parade dynamic temporal geste
for a time sequence. Unlike feed forward neural networks, RNNs can use their internal state( memory) to reuse sequences of inputs. 5. Multilayer intermittent network a. In this type of network, processing element affair can be directed to the processing element in the same subcaste and in the antedating subcaste forming a multilayer intermittent network. b. They perform the same task for every element of a sequence, with the affair being depended on the former calculations. Inputs are not demanded at each time step. c. The main point of a multilayer intermittent neural network is its retired state, which captures information about a sequence. Que4.6. bandy the benefits of artificial neural network. Answer 1. Artificial neural networks are flexible and adaptive. 2. Artificial neural networks are used in sequence and pattern recognition systems, data processing, robotics, modeling,etc. 3. ANN acquires knowledge from their surroundings by conforming to internal and external parameters and they break complex problems which are delicate to manage. 4. It generalizes knowledge to produce acceptable responses to unknown situations. 5. Artificial neural networks are flexible and have the capability to learn, generalize and adapts to situations grounded on its findings. 6. This function allows the network to efficiently acquire knowledge by literacy. This is a distinct advantage over a traditionally direct network that is shy when it comes to modellingnon-linear data. 7. An artificial neuron network is able of lesser fault forbearance than a traditional network. Without the loss of stored data, the network is suitable to regenerate a fault in any of its factors. 8. An artificial neuron network is grounded on adaptive literacy. Que4.7. Write short note on grade descent. Answer 1. grade descent is an optimization algorithm used to minimize some function by iteratively moving in the direction of steepest descent as defined by the negative of the grade. 2. A grade is the pitch of a function, the degree of change of a parameter with the quantum of change in another parameter. 3. Mathematically, it can be described as the partial derivations of a set of parameters with respect to its inputs. The more the grade, the steeper the pitch. 4. grade Descent is a convex function. 5. grade Descent can be described as an iterative system which is used to find the values of the parameters of a function that minimizes the cost function as much as possible. 6. The parameters are originally defined a particular value and from that, grade Descent run in an iterative fashion to find the optimal values of the parameters, using math, to find the minimal possible value of the given cost function.

Leave a Comment