In this paper, we introduce a distributed version of the classical stochastic Multi-Arm Bandit (MAB) problem. Our setting consists of a large number of agents $n$ that collaboratively and simultaneously solve the same instance of $K$ armed MAB to minimize the average cumulative regret over all agents. The agents can communicate and collaborate among each other \emph{only} through a pairwise asynchronous gossip based protocol that exchange a limited number of bits. In our model, agents at each point decide on (i) which arm to play, (ii) whether to, and if so (iii) what and whom to communicate with. Agents in our model are decentralized, namely their actions only depend on their observed history in the past. We develop a novel algorithm in which agents, whenever they choose, communicate only arm-ids and not samples, with another agent chosen uniformly and independently at random. The per-agent regret scaling achieved by our algorithm is $O \left( \frac{\lceil\frac{K}{n}\rceil+\log(n)}{\Delta} \log(T) + \frac{\log^3(n) \log \log(n)}{\Delta^2} \right)$. Furthermore, any agent in our algorithm communicates only a total of $\Theta(\log(T))$ times over a time interval of $T$. We compare our results to two benchmarks - one where there is no communication among agents and one corresponding to complete interaction. We show both theoretically and empirically, that our algorithm experiences a significant reduction both in per-agent regret when compared to the case when agents do not collaborate and in communication complexity when compared to the full interaction setting which requires $T$ communication attempts by an agent over $T$ arm pulls.

In this paper, we study the problem of computing $U$-statistics of degree $2$, i.e., quantities that come in the form of averages over pairs of data points, in the local model of differential privacy (LDP). The class of $U$-statistics covers many statistical estimates of interest, including Gini mean difference, Kendall's tau coefficient and Area under the ROC Curve (AUC), as well as empirical risk measures for machine learning problems such as ranking, clustering and metric learning. We first introduce an LDP protocol based on quantizing the data into bins and applying randomized response, which guarantees an $\epsilon$-LDP estimate with a Mean Squared Error (MSE) of $O(1/\sqrt{n}\epsilon)$ under regularity assumptions on the $U$-statistic or the data distribution. We then propose a specialized protocol for AUC based on a novel use of hierarchical histograms that achieves MSE of $O(\alpha^3/n\epsilon^2)$ for arbitrary data distribution. We also show that 2-party secure computation allows to design a protocol with MSE of $O(1/n\epsilon^2)$, without any assumption on the kernel function or data distribution and with total communication linear in the number of users $n$. Finally, we evaluate the performance of our protocols through experiments on synthetic and real datasets.

Recently there has been a flurry of interest around using pipeline parallelism while training neural networks. Pipeline parallelism enables larger models to be partitioned spatially across chips and within a chip, leading to both lower network communication and overall higher hardware utilization. Unfortunately, to preserve statistical efficiency, existing pipeline-parallelism techniques sacrifice hardware efficiency by introducing bubbles into the pipeline and/or incurring extra memory costs. In this paper, we investigate to what extent these sacrifices are necessary. Theoretically, we derive a simple but robust training method, called PipeMare, that tolerates asynchronous updates during pipeline-parallel execution. Using this, we show empirically, on a ResNet network and a Transformer network, that PipeMare can achieve final model qualities that match those of synchronous training techniques (at most 0.9% worse test accuracy and 0.3 better test BLEU score) while either using up to 2.0X less weight and optimizer memory or being up to 3.3X faster than other pipeline parallel training techniques. To the best of our knowledge we are the first to explore these techniques and fine-grained pipeline parallelism (e.g. the number of pipeline stages equals to the number of layers) during neural network training.

In exchange for large quantities of data and processing power, deep neural networks have yielded models that provide state of the art predication capabilities in many fields. However, a lack of strong guarantees on their behaviour have raised concerns over their use in safety-critical applications. A first step to understanding these networks is to develop alternate representations that allow for further analysis. It has been shown that neural networks with piecewise affine activation functions are themselves piecewise affine, with their domains consisting of a vast number of linear regions. So far, the research on this topic has focused on counting the number of linear regions, rather than obtaining explicit piecewise affine representations. This work presents a novel algorithm that can compute the piecewise affine form of any fully connected neural network with rectified linear unit activations.

We propose a new forecasting method for predicting load demand and generation scheduling. Accurate week-long forecasting of load demand and optimal power generation is critical for efficient operation of power grid systems. In this work, we use a synthetic data set describing a power grid with 700 buses and 134 generators over a 365-days period with data synthetically generated at an hourly rate. The proposed approach for week-long forecasting is based on the Gaussian process regression (GPR) method, with prior covariance matrices of the quantities of interest (QoI) computed from ensembles formed by up to twenty preceding weeks of QoI observations. Then, we use these covariances within the GPR framework to forecast the QoIs for the following week. We demonstrate that the the proposed ensemble GPR (EGPR) method is capable of accurately forecasting weekly total load demand and power generation profiles. The EGPR method is shown to outperform traditional forecasting methods including the standard GPR and autoregressive integrated moving average (ARIMA) methods.

The next frontier in solar system exploration will be missions targeting extreme and rugged environments such as caves, canyons, cliffs and crater rims of the Moon, Mars and icy moons. These environments are time capsules into early formation of the solar system and will provide vital clues of how our early solar system gave way to the current planets and moons. These sites will also provide vital clues to the past and present habitability of these environments. Current landers and rovers are unable to access these areas of high interest due to limitations in precision landing techniques, need for large and sophisticated science instruments and a mission assurance and operations culture where risks are minimized at all costs. Our past work has shown the advantages of using multiple spherical hopping robots called SphereX for exploring these extreme environments. Our previous work was based on performing exploration with a human-designed baseline design of a SphereX robot. However, the design of SphereX is a complex task that involves a large number of design variables and multiple engineering disciplines. In this work we propose to use Automated Multidisciplinary Design and Control Optimization (AMDCO) techniques to find near optimal design solutions in terms of mass, volume, power, and control for SphereX for different mission scenarios.

Launched in January this year, the Maha Agri Tech project seeks to use technology to address various cultivation risks ranging from poor rains to pest attacks, accurately predict crop-wise and area-wise yield and eventually to use this data to inform policy decisions including pricing, warehousing and crop insurance. When farmers in six districts of Maharashtra begin sowing for the coming rabi season, this project will enter its second phase where artificial intelligence and satellite imagery will be used to mitigate risks. Fields of the farmers that are part of the project will be monitored via satellite images at every stage right until the harvest. In its first phase the Maha Agri Tech project used satellite images and analysis from the Maharashtra Remote Sensing Application Centre (MRSAC) and the National Remote Sensing Centre (NRSC) in Hyderabad to assess the acreage and the conditions of select crops in select talukas. In its second phase, various data sets from diverse data providers will be combined to build yield modelling and a geospatial database of soil nutrients, rainfall, moisture stress and other parameters to facilitate location-specific advisories to farmers.

IBM released a report in September about which I have such mixed feelings. On the one hand, I feel excited about the possibilities, and on the other, I am alarmed and trying to understand the implications. But let's take a step back... According to IBM, up to 120 million workers across 12 of the largest economies in the world will need to be retrained or deskilled due to the integration of AI or intelligent automation. This statistic in itself is somewhat alarming, but it is compounded by the fact that only 41% of the CEOs surveyed by IBM says they have the people, skills or resources in place in order to execute their business strategies.

Shell is to use machine learning for the management of hybrid domestic heat pumps in a new partnership with PassivSystems. The B-Snug system for domestic customers uses machine learning to continuously monitor the temperature in the home and analyses weather forecasts to automatically switch between an air source heat pump and traditional boiler. The technology platform is designed so that the heat pump is prioritised whenever possible, PassivSystems said. Due to the prioritisation of heat from an electric source, the initiative is set to reduce carbon emissions from domestic properties. Colin Calder, CEO at PassivSystems, said there is a "clear message" from the Committee on Climate Change (CCC) that homes must be decarbonised for net zero and that the launch of the hybrid heating system directly contributes to the CCC's goal of deploying 10 million hybrids by 2035.

Employee activism and outside pressure have pushed big tech companies like Amazon, Microsoft and Google into promising to slash their carbon emissions. When Microsoft held an all-staff meeting in September, an employee asked CEO Satya Nadella if it was ethical for the company to be selling its cloud computing services to fossil fuel companies, according to two other Microsoft employees who described the exchange on condition they not be named. Such partnerships, the worker told Nadella, were accelerating the oil companies' greenhouse gas emissions. Microsoft and other tech giants have been competing with one another to strike lucrative partnerships with ExxonMobil, Chevron, Shell, BP and other energy firms, in many cases supplying them not just with remote data storage but also artificial intelligence tools for pinpointing better drilling spots or speeding up refinery production. The oil and gas industry is spending roughly $20 billion each year on cloud services, which accounts for about 10% of the total cloud market, according to Vivek Chidambaram, a managing director of Accenture's energy consultancy.