• Sai Harshini Garapati
• LeelaMadhav Somu

Deep learning is an important tool for many tasks, but it can be hard to teach models that are good at the generalization of new data. We are proposing a new approach to the training of advanced learning models, which we call target space training in this paper. The model is taught to reduce the loss function of the target space, as opposed to input space, during training for Target Space. This is useful in addressing the problem of exploding gradients, which makes a model less sensitive to noise from training data. In particular, we tested the performance of targeted space training on various tasks such as image classification, natural language processing, and speech recognition. We have found that training in target space can improve the generalization of deep learning models for these tasks. The advantages and disadvantages of targeting space training have also been explored. Target space training may be more computationally expensive than traditional training methods, but it may also improve the generalization of the model. We believe that target space training is a promising new approach to training deep learning models, and we plan to continue our research in this area. The advantages and disadvantages of targeting space training have also been explored. Target space training may be more computationally expensive than traditional training methods, but it may also improve the generalization of the model. We believe that target space training is a promising new approach to training deep learning models, and we plan to continue our research in this area.”

DeepLearning, Target Space Training, Exploding Gradients, NLP.

"Deep Learning at Target Space for Brain Tumor Analysis and Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 1, page no.g286-g294, January-2023, Available :http://www.jetir.org/papers/JETIR2301638.pdf

"Deep Learning at Target Space for Brain Tumor Analysis and Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 1, page no. ppg286-g294, January-2023, Available at : http://www.jetir.org/papers/JETIR2301638.pdf