Shivkumar Kalyanaraman’s research while affiliated with Microsoft and other places

... For example, Y. Al-Wreikat et al. [1] propose to partition drivers into multiple classes and quantify the average VECs. The data-driven methods, on the other hand, aim to automatically extract and utilize relevant knowledge by leveraging machine learning tools, e.g., linear regression (LR) [7], decision tree (DT) [25], support vector machine (SVM) [21], etc. Inspired by the recent advances of deep learning, deep neural networks such as Long-Short-Term Memory (LSTM) [4] and Transformer [31] have also been adopted to analyze road-level energy consumption. However, existing approaches make predictions based on handcrafted statistical features, which overlook the personalized nature of varying driving behaviors under different travel contexts. ...

Reference:

A Preference-aware Meta-optimization Framework for Personalized Vehicle Energy Consumption Estimation

... In solar energy systems, machine learning algorithms enhance solar panel performance, increase energy forecasting, and optimize energy storage systems. For instance, machine-learning techniques have been used to detect and localize solar panel faults, drastically reducing the time required to identify and rectify faulty cells (Ahan et al., 2021). In addition, machine learning-based solar energy forecasting systems have demonstrated high accuracy (Partheeban et al., 2022). ...

Reference:

Machine learning and the renewable energy revolution: Exploring solar and wind energy solutions for a sustainable future including innovations in energy storage

... In this case, the according companies might be able to shift tasks among data centers depending on the momentary energy costs at both data centers [33], [91]. Moreover, also the edge data centers built currently may contribute to flexible demand [92]. ...

Reference:

A Structural Analysis of the User Behavior Dynamics for Environmentally Sustainable ICT

... Notwithstanding the significant advancements achieved in modeling methodology and computational capability, modern climate models persistently encounter a myriad of challenges. This is illustrated, for instance, by the requirement for precise models of complex physical processes, the integration of a vast array of data types, and predictions with an actionable resolution for the planning and executing of local decisions [31]. A considerable number of the presently operational models rely on statistical techniques, which while beneficial, do not always account for the omission of precise scientific data about the climate system. ...

Reference:

Advancing Climate Forecasting Accuracy Through Neurosymbolic Integration: Unveiling the Neurosymbolic Climate Inference and Prediction (NS-CIP) Model’s Approach to Predicting Extreme Weather Events

... Random Forest is one of the most widely used Classification algorithms, which uses the Ensemble Learning Technique [32]. One of the main advantages of Random Forest is that it creates many decision trees on the input subset data and combines all the outputs into a final prediction as shown in Figure 5 [33]. Random Forest methods are less prone to overfitting problems and tend to have better accuracy than other classification methods. ...

Reference:

Data-Driven Soil Analysis and Evaluation for Smart Farming Using Machine Learning Approaches

... Many studies compared the satellite SM observations with the available in-situ data over the Indian region (Bhimala and Goswami 2015;Das et al. 2018;Bhimala and Rakesh 2019) and noticed that the two data sets are generally well consistent with high (moderate) correlation during wet (dry) season. In recent years, emphasize is given to the satellite-derived SM and also given importance of the inclusion of SM initial information in the NWP. ...

Reference:

Influence of ASCAT soil moisture on prediction of track and intensity of landfall tropical cyclones

... Artificial Neural Networks ANN have been shown by Wojdyga [17] to be an effective approach to analyze data from previous heating seasons. Moreover, there has been a clear increase in interest in the use of machine learning methods, such as support vector machines (SVM) [18], random forest [19], deep learning [20] and gradient boosting [21]. Additionally, a reinforcement machine learning technique was implemented to optimize the combustion control strategy [14]. ...

Reference:

Industrial Application of Machine Learning Clustering for a Combined Heat and Power Plant: A Pavlodar Case Study

... Using this intelligent automation, not only cost parameter is worked upon but the overall efficiency of the device is also improved. To boost up energy yield of the solar PVC by efficient photonic energy harvesting, the use of IoT has proven useful [29,30]. The node MCU, a minicomputer, is used to implement IoT in the proposed system. ...

Reference:

Internet of Things Driven Framework for Smart Solar Energy System

... For example, the efficiency of PV panels significantly drops during extreme heat. Cloud and aerosols result in very low power production by reducing near-surface solar radiation 11 . In Europe, it was found that all regions have experienced periods of very low solar power over the past 23 years (1995-2017), though the severity and driven weather patterns differ 12 . ...

Reference:

Photovoltaic installations are extensively deployed in areas at risk of extremely low production

... In this work, we present a novel analytics approach [7] (Fig. 1) to estimate the connectivity model of a radial 1 distribution network. Our techniques are novel as they are purely based upon a time series of power measurements collected by various meters in the distribution grid. ...

Reference:

Inferring connectivity model from meter measurements in distribution networks