New Delhi, Jul 02: Bio-derived fuels are gaining widespread attention among the scientific community across the world. The work on biofuels is in response to the global call for reducing carbon emissions associated with the use of fossil fuels. In India too, biofuels have caught the imagination of researchers. For instance, researchers of the Indian Institute of Technology (IIT) Hyderabad have started using computational methods to understand the factors and impediments in incorporating biofuels into the fuel sector in India. A unique feature of this work is that the framework considers revenue generation not only as an outcome of sales of the biofuel but also in terms of carbon credits via greenhouse gas emission savings throughout the project lifecycle.

IIT Hyderabad Researchers are using computational methods to understand the factors and impediments in incorporating biofuels into the fuel sector in India. This work has been spurred by the increasing need to replace fossil fuels by bio-derived fuels, which, in turn, is driven by the dwindling fossil fuel reserves all over the world, and pollution issues associated with the use of fossil fuels. The model developed by the IIT Hyderabad team has shown that in the area of bioethanol integration into mainstream fuel use, the production cost is the highest (43 per cent) followed by import (25 per cent), transport (17 per cent), infrastructure (15 per cent) and inventory (0.43 per cent) costs. The model has also shown that feed availability to the tune of at least 40 per cent of the capacity is needed to meet the projected demands. A unique feature of this work is that the framework considers revenue generation not only as an outcome of sales of the biofuel but also in terms of carbon credits via greenhouse gas emission savings throughout the project lifecycle.

Resolving the exploration-exploitation trade-off remains a fundamental problem in the design and implementation of reinforcement learning (RL) algorithms. In this paper, we focus on model-free RL using the epsilon-greedy exploration policy, which despite its simplicity, remains one of the most frequently used forms of exploration. However, a key limitation of this policy is the specification of $\varepsilon$. In this paper, we provide a novel Bayesian perspective of $\varepsilon$ as a measure of the uniformity of the Q-value function. We introduce a closed-form Bayesian model update based on Bayesian model combination (BMC), based on this new perspective, which allows us to adapt $\varepsilon$ using experiences from the environment in constant time with monotone convergence guarantees. We demonstrate that our proposed algorithm, $\varepsilon$-\texttt{BMC}, efficiently balances exploration and exploitation on different problems, performing comparably or outperforming the best tuned fixed annealing schedules and an alternative data-dependent $\varepsilon$ adaptation scheme proposed in the literature.

Non-uniform sampling arises when an experimenter does not have full control over the sampling characteristics of the process under investigation. Moreover, it is introduced intentionally in algorithms such as Bayesian optimization and compressive sensing. We argue that Stochastic Differential Equations (SDEs) are especially well-suited for characterizing second order moments of such time series. We introduce new initial estimates for the numerical optimization of the likelihood, based on incremental estimation and initialization from autoregressive models. Furthermore, we introduce model truncation as a purely data-driven method to reduce the order of the estimated model based on the SDE likelihood. We show the increased accuracy achieved with the new estimator in simulation experiments, covering all challenging circumstances that may be encountered in characterizing a non-uniformly sampled time series. Finally, we apply the new estimator to experimental rainfall variability data.

Generating high quality uncertainty estimates for sequential regression, particularly deep recurrent networks, remains a challenging and open problem. Existing approaches often make restrictive assumptions (such as stationarity) yet still perform poorly in practice, particularly in presence of real world non-stationary signals and drift. This paper describes a flexible method that can generate symmetric and asymmetric uncertainty estimates, makes no assumptions about stationarity, and outperforms competitive baselines on both drift and non drift scenarios. This work helps make sequential regression more effective and practical for use in real-world applications, and is a powerful new addition to the modeling toolbox for sequential uncertainty quantification in general.

Determination of inspection and maintenance policies for minimizing long-term risks and costs in deteriorating engineering environments constitutes a complex optimization problem. Major computational challenges include the (i) curse of dimensionality, due to exponential scaling of state/action set cardinalities with the number of components; (ii) curse of history, related to exponentially growing decision-trees with the number of decision-steps; (iii) presence of state uncertainties, induced by inherent environment stochasticity and variability of inspection/monitoring measurements; (iv) presence of constraints, pertaining to stochastic long-term limitations, due to resource scarcity and other infeasible/undesirable system responses. In this work, these challenges are addressed within a joint framework of constrained Partially Observable Markov Decision Processes (POMDP) and multi-agent Deep Reinforcement Learning (DRL). POMDPs optimally tackle (ii)-(iii), combining stochastic dynamic programming with Bayesian inference principles. Multi-agent DRL addresses (i), through deep function parametrizations and decentralized control assumptions. Challenge (iv) is herein handled through proper state augmentation and Lagrangian relaxation, with emphasis on life-cycle risk-based constraints and budget limitations. The underlying algorithmic steps are provided, and the proposed framework is found to outperform well-established policy baselines and facilitate adept prescription of inspection and intervention actions, in cases where decisions must be made in the most resource- and risk-aware manner.

Earth observation (EO) by airborne and satellite remote sensing and in-situ observations play a fundamental role in monitoring our planet. In the last decade, machine learning has attained outstanding results in the estimation of bio-geo-physical variables from the acquired images at local and global scales in a time-resolved manner. Gaussian processes (GPs) [1], as flexible nonparametric models to find functional relationships, have excelled in EO problems in recent years, mainly introduced for model inversion and emulation of complex codes [2]. GPs provide not only accurate estimates but also principled uncertainty estimates for the predictions. Besides, GPs can easily accommodate multimodal data coming from different sensors and from multitemporal acquisitions. Due to their solid Bayesian formalism, GPs can include prior physical knowledge about the problem, and allow for a formal treatment of uncertainty quantification and error propagation. In remote sensing, we often deal with radiative transfer models (RTMs) which implement the equations of energy transfer. These codes are needed for modelling, understanding, and predicting some variables of interest related to the state of the land cover, water bodies and atmosphere.

Twice a year, the world's fastest supercomputers take a test to see which is top of class. These hundred-million-dollar machines usually run on hundreds of thousands of processors, occupy warehouse floors, gobble up copious amounts of energy, and crunch numbers at an ungodly pace. All that computing is directed at some of humanity's toughest challenges with the likes of advanced climate modeling or protein simulations to help cure diseases. For the last two years, the US's Summit was the fastest supercomputer on the planet. But this week, a new system took the crown.

Digitalisation has already had a major impact on the world of business, but now a project helmed by Deloitte has helped the World Wildlife Fund wield new technology in the fight to protect the world's most fragile ecosystems. Supported by the Deloitte Impact Foundation, the Cognitive Deforestation Prevention programme uses artificial intelligence to prevent illegal deforestation. According to Mark Boersma, the Senior Manager within Consulting who leads the Deloitte Impact Foundation initiative, the ideas behind the Cognitive Deforestation Prevention began to form five years ago, during a trip to the Indonesian island of Sumatra. During the expedition with his wife, they encountered rare species such as orang-utans and a group of black gibbons, but they were also struck with the serious toll human activity was taking on the shrinking natural world. Recounting the experience on Deloitte's website, Boersma said, "I remember that around the rainforest, we were struck with acres and acres of palm oil trees which had often come in place for the less intrusive rubber plantations. This is when I started to think about the effect that our palm oil consumption is having on our world… [When] WWF sent out an open request for proposals for their Early Warning System and I instantly knew that I wanted to be a part of it."

Learned neural solvers have successfully been used to solve combinatorial optimization and decision problems. More general counting variants of these problems, however, are still largely solved with hand-crafted solvers. To bridge this gap, we introduce belief propagation neural networks (BPNNs), a class of parameterized operators that operate on factor graphs and generalize Belief Propagation (BP). In its strictest form, a BPNN layer (BPNN-D) is a learned iterative operator that provably maintains many of the desirable properties of BP for any choice of the parameters. Empirically, we show that by training BPNN-D learns to perform the task better than the original BP: it converges 1.7x faster on Ising models while providing tighter bounds. On challenging model counting problems, BPNNs compute estimates 100's of times faster than state-of-the-art handcrafted methods, while returning an estimate of comparable quality.