Human Activity Recognition has been a favorite topic for the scholars not only because of its wid... more Human Activity Recognition has been a favorite topic for the scholars not only because of its wide scale acceptance in the industry but areas which may help in medical and in our normal household works as well. Since to make this technology available to the last person standing in the queue it is important that models compiled and trained in this field are not just high performing but optimized as such with incurs the least overhead. And thus bringing TinyML into the picture which has specialty in the field of optimizing the model w.r.t. the size of the model, energy consumption, network bandwidth usage etc. Thus this work includes using optimizing techniques such as pruning and quantization on the pre-proposed models and analyze the changes it causes in such models w.r.t accuracy and size. Our work is able infer that by using both Pruning and Quantization techniques on a human activity recognition model we can compress a model up to 10 time without hampering severe diversion to the accuracy of the model. We have taken three models and UCI-HAR dataset and compare the outcomes of the experiment.
Wearable Device Design for Cattle Behavior Classification Using IoT and Machine Learning
Advances in intelligent systems and computing, 2021
The term Internet of Things refers to the huge network of interconnected smart devices (sensors, ... more The term Internet of Things refers to the huge network of interconnected smart devices (sensors, actuators, RFID tags and readers) that can share and communicate information without any human intervention. One of the diverse applications of IoT is in the field of dairy farming for cattle behavior classification, automated heat detection (time period when the cows are sexually receptive) and calving time prediction. In this paper, first the work proposed by various researchers for cattle behavior classification using supervised and unsupervised machine learning techniques has been summarized. One of the limitations found in previous work was that the researchers used LM35 sensor for cow’s body temperature measurement which read environmental temperature readings thus giving inaccurate results. Second limitation was quick discharge of wearable device due to high power consumption modules like Wi-Fi module. The proposed model discussed in this paper overcomes these limitations. The sensor system is divided into two parts: wearable transmitter module and receiver module. The wearable collar transmitter module consisted of only sensors and wireless transceiver module. It uses contactless infrared temperature sensor (Tmp006) instead of LM35. The high-power consumption modules were placed in the receiver module which had continuous power supply. Hence, the power requirements were minimized and the battery life was increased in the wearable from few hours to 4 days.
A Transformers Approach to Detect Depression in Social Media
The invention of diverse instant messaging applications or social media platforms has empowered e... more The invention of diverse instant messaging applications or social media platforms has empowered everybody to effortlessly create, widely distribute and convey their views, feelings, innermost thoughts, fears in addition to achievements, with the whole world. With the upgradation of technology like Internet of Things and Artificial Intelligence which are now occupying every single aspect of our human lives. They have made it possible to connect all our devices to each other and operate them with just a single touch and likewise with similar ease, anyone can share and express their opinions, be it in the form of long messages or poems or images on social media platforms like Reddit, Telegram, Facebook, Instagram, Twitter, WhatsApp, etc. This exchange of ideas, views and keeping in touch with each other became all the more important especially in times of Corona when the pandemic forced everyone to stay inside their homes and function from their comfortable environments. The problem of depression became more pronounced in these lockdown times and exacerbated the loneliness of those suffering from mental health disorders. Therefore, early detection of depression through the use of social media information via deep learning techniques can phenomenally revolutionise the area of depression detection, where most of the people took to social media to display their feelings. This paper implements certain baseline models such as Support Vector Machines, Linear Classifiers etc. and the Transformers model on Reddit dataset to thereby, achieve a higher accuracy in detecting depression in social media users.
Human Activity Recognition has been a favorite topic for the scholars not only because of its wid... more Human Activity Recognition has been a favorite topic for the scholars not only because of its wide scale acceptance in the industry but areas which may help in medical and in our normal household works as well. Since to make this technology available to the last person standing in the queue it is important that models compiled and trained in this field are not just high performing but optimized as such with incurs the least overhead. And thus bringing TinyML into the picture which has specialty in the field of optimizing the model w.r.t. the size of the model, energy consumption, network bandwidth usage etc. Thus this work includes using optimizing techniques such as pruning and quantization on the pre-proposed models and analyze the changes it causes in such models w.r.t accuracy and size. Our work is able infer that by using both Pruning and Quantization techniques on a human activity recognition model we can compress a model up to 10 time without hampering severe diversion to the...
A Transformers Approach to Detect Depression in Social Media
2021 International Conference on Artificial Intelligence and Smart Systems (ICAIS), 2021
The invention of diverse instant messaging applications or social media platforms has empowered e... more The invention of diverse instant messaging applications or social media platforms has empowered everybody to effortlessly create, widely distribute and convey their views, feelings, innermost thoughts, fears in addition to achievements, with the whole world. With the upgradation of technology like Internet of Things and Artificial Intelligence which are now occupying every single aspect of our human lives. They have made it possible to connect all our devices to each other and operate them with just a single touch and likewise with similar ease, anyone can share and express their opinions, be it in the form of long messages or poems or images on social media platforms like Reddit, Telegram, Facebook, Instagram, Twitter, WhatsApp, etc. This exchange of ideas, views and keeping in touch with each other became all the more important especially in times of Corona when the pandemic forced everyone to stay inside their homes and function from their comfortable environments. The problem of depression became more pronounced in these lockdown times and exacerbated the loneliness of those suffering from mental health disorders. Therefore, early detection of depression through the use of social media information via deep learning techniques can phenomenally revolutionise the area of depression detection, where most of the people took to social media to display their feelings. This paper implements certain baseline models such as Support Vector Machines, Linear Classifiers etc. and the Transformers model on Reddit dataset to thereby, achieve a higher accuracy in detecting depression in social media users.
Wearable Device Design for Cattle Behavior Classification Using IoT and Machine Learning
Advances in Intelligent Systems and Computing, 2021
The term Internet of Things refers to the huge network of interconnected smart devices (sensors, ... more The term Internet of Things refers to the huge network of interconnected smart devices (sensors, actuators, RFID tags and readers) that can share and communicate information without any human intervention. One of the diverse applications of IoT is in the field of dairy farming for cattle behavior classification, automated heat detection (time period when the cows are sexually receptive) and calving time prediction. In this paper, first the work proposed by various researchers for cattle behavior classification using supervised and unsupervised machine learning techniques has been summarized. One of the limitations found in previous work was that the researchers used LM35 sensor for cow’s body temperature measurement which read environmental temperature readings thus giving inaccurate results. Second limitation was quick discharge of wearable device due to high power consumption modules like Wi-Fi module. The proposed model discussed in this paper overcomes these limitations. The sen...
Uploads
Papers by Bholanath Roy