Artificial Intelligence for Balance Control and Fall Detection of Elderly People

    Balance control in elderly people is one of the key issues of old age. Artificial Intelligence can play a big role to solve this issue. In this research work, we demonstrate the application of machine learning techniques for posture alignments and the control of the body center of mass for disabled people.

    Designing automated balance control system for elderly people is one of the key projects of our Compassionate AI Lab. Here, Dr. Amit Ray discuses about one of the recent projects of AI using deep learning algorithms for automatic balance control of elderly people. He explains how machine learning algorithms can be used to study and improve the dynamical properties of postural stability of elderly people. The project focuses on how image recognition, human-body joint dynamics, and path navigation methods of artificial intelligence can be used  to eliminate the imbalance, fall and injury of elderly people or for physically challenged people.

    AI for Balance-Control Fall Detection of Elderly People

    Compassionate Artificial Intelligence can be used for helping elderly people in many ways. Here, we discuss about one of our recent project of using AI & deep learning techniques for automatic balance control. The machine learning algorithms are used to improve dynamical properties of postural stability. In this project AI based machine learning algorithms are used to find the insights into the person specific postural strategies for older adults in order to adapt to the postural challenges during sleeping, standing, turning and walking. To study the body movement behavior of elderly people accurately, it is necessary to observe and record their movement trajectory and joint movements quantitatively and precisely in three dimensions.

    Read more ..

    7 Limitations of Deep Learning Algorithms of AI

    7 Limitations of Deep Learning Algorithms of AI

    Sri Amit Ray tells us about the power and limitations of Deep Learning Architectures of Artificial Intelligence (AI).

    Deep learning architectures of Artificial Intelligence has provided remarkable capabilities and advances in voice recognition, face recognition, pattern recognition, image understanding, natural language processing, game planning language translation, and search engine optimization. Deep learning is the key technology behind self-driving car. However, deep learning algorithms of AI have several inbuilt limitations. This article is focused to explain the power and limitations of current deep learning algorithms. It also provides the directions and lights to overcome the limits of deep learning algorithms to achieve higher levels learning capabilities.

    Deep Learning Algorithms in AI Power and Limitations

    Types of Machine Learning:

    There are three core types of machine learning: supervised learning, unsupervised learning, and reinforcement learning. In supervised learning, the algorithm uses the training data to learn a link between the input and the outputs. It predicts the output from the trained network.  

    Unsupervised learning does not use output data. It uses dimensionality reduction algorithms, Clustering algorithm, such as K-means etc. The most common unsupervised learning method is cluster analysis. Unsupervised learning algorithms can perform more complex processing tasks than supervised learning systems.

     

    Reinforcement learning algorithms attempt to find the best ways to earn the greatest reward. Rewards can be winning a game, earning more points or beating other opponents. Given its and the environment’s states, the agent will choose the action which will maximize its reward or will explore a new possibility. Reinforcement learning is useful in cases where the solution space is big.

     

    What is deep learning?

    Deep learning is a sub-field of artificial intelligence. It attempts to mimic the layers of neurons in the brain’s neocortex. Today deep learning algorithms are the heart of designing intelligent systems. During the training phase, deep learning architectures learns to discover the underline hidden patterns from the data and it creates multiple layers of automatic data abstraction. Deep learning algorithms follows non-linearity, distributed representation, parallel computation, adaptive, and self-organizing architectures. 

    A deep learning neural network architecture is an artificial neural network (ANN) with multiple hidden layers between the input and output layers. Deep learning is also known as deep structured learning or hierarchical learning. Learning can be supervised, semi-supervised or unsupervised. Deep reinforcement learning algorithms are applied for learning to play video games, and robotics, allowing control policies for robots to be learned directly from camera inputs in the real world.  They are used as deep neural networks, deep belief networks and recurrent neural networks.

    In simplistic term deep learning employs an algorithm that adjusts the mathematical weights between nodes, so that an input leads to the right output through the hidden layers. The main difference between shallow learning and deep learning is; shallow learning neural network architectures have only one hidden layer but deep learning neural networks have many hidden layers and Deep learning includes both supervised and unsupervised learning.  

    Power of Deep Learning:

    With the advancement of computing power and graphics processing units (GPUs), and deep learning algorithm it is now possible to train large scale complex AI models that enable deep insights of real life complex problems. It has the hidden power to apply in almost every field of life.   The greatest advantage of deep learning is that it hides the complexity from the developers. There is a plethora of deep learning libraries, algorithms and capabilities. They are also available in popular open-source frameworks. Deep learning algorithms has also given tremendous capabilities to integrate artificial intelligence with emerging technologies like; Internet of Things (IoT), Big data, Drone, brain-computer-interface etc.

    Deep Recurrent Learning:

    A recurrent neural network (RNN) is a class of artificial neural network where connections between units form a directed graph along a sequence. This allows it to exhibit dynamic temporal behavior for a time sequence. Unlike feedforward neural networks, RNNs can use their internal state (memory) to process sequences. 

    The book Compassionate Artificial Superintelligence AI 5.0,  discussed the power, scope and challenges of deep learning algorithms in details.   

    Limitations of Deep Learning:

    Deep learning is remarkably powerful for solving classification problems but all problems can not be represented in classification format. Some of the limitations of common deep learning algorithms are as follows:

    1. Lacks common sense. Common sense is the practice of acting intelligently in everyday situations. It is the ability to draw conclusions even with limited experience. Deep learning algorithms can not draw conclusions in the cross-domain boundary areas. 
    2. Lacks understandings about exact underlines laws of the input data.  On the basis of training network and data, we can only estimate the output but cannot make a claim that it would be exactly 100 %. Here only approximations are made.

    3. Lacks general intelligence and multiple domain knowledge integration. The intelligence of human civilization accelerates due to connectivity between people. Neural networks fed inaccurate or incomplete data will produce the wrong results. The outcomes can be embarrassing. 
    4. Unable to learn from limited examples. It’s intelligence mostly depends on the training dataset have been used. It cannot be used in problems that dynamically change. 
    5. Less powerful beyond classification problems.  Most Deep Learning algorithms seem to focus on classification or dimensional reduction. They are less powerful for long-term planning. It lacks creativity and imagination. 
    6. Lack of global generalization. Human can imagine and anticipate different possible problem cases, and provides solutions and perform long-term planning for that. 
    7. Deep learning is certainly limited in its current form, because almost all the successful applications of it use supervised learning with human-annotated data. It cannot take complex decisions beyond any previous training. However, Deep Q learning algorithms are small steps towards that. 

    Conclusions:

    Deep learning is one of the most powerful tool of AI. However, it has many limitations. We elaborate the seven limitations of deep learning algorithms of artificial intelligent systems. Serving humanity intelligently, held up as the “gold standard” of AI based systems and deep learning architecture is one of the most powerful tools for that. However, it alone is not sufficient. It needs many combinations with other technologies.

    Sources:

    1. Compassionate Superintelligence (AI-5.0) By Dr. Amit Ray, Inner Light Publishers, 2018. 
    2. Deep Learning Review, Nature, Vol. 521, By Yann LeCun et al.
    Read more ..


Contact us | About us | Privacy Policy and Terms of Use |

Copyright ©AmitRay.com, 2010-2024, All rights reserved. Not to be reproduced.