Abstract

In classification models fairness can be ensured by solving a constrained optimization problem. We focus on fairness constraints like Disparate Impact, Demographic Parity, and Equalized Odds, which are non-decomposable and non-convex. Researchers define convex surrogates of the constraints and then apply convex optimization frameworks to obtain fair classifiers. Surrogates serve only as an upper bound to the actual constraints, and convexifying fairness constraints might be challenging. We propose a neural network-based framework, FNNC, to achieve fairness while maintaining high accuracy in classification. The above fairness constraints are included in the loss using Lagrangian multipliers. We prove bounds on generalization errors for the constrained losses which asymptotically go to zero. The network is optimized using two-step mini-batch stochastic gradient descent. Our experiments show that FNNC performs as good as the state of the art, if not better. The experimental evidence supplements our theoretical guarantees. In summary, we have an automated solution to achieve fairness in classification, which is easily extendable to many fairness constraints.

Abstract

Around 40% of global energy produced is consumed by buildings. By using renewable energy resources we can alleviate the dependence on electrical grids. Recent trends focus on incentivizing consumers to reduce their demand consumption during peak hours for sustainable demand response. To minimize the loss, the distributor companies should target the right set of consumers and demand the right amount of electricity reductions.This paper proposes a novel bounded integer min-knapsack algorithm and shows that the algorithm, while allowing for multiple unit reduction, also optimizes the loss to the distributor company within a factor of two (multiplicative) and a problem dependent additive constant. Existing CMAB algorithms fail to work in this setting due to nonmonotonicity of reward function and time varying optimal sets. We propose a novel algorithm (TwinMinKPDR-CB) to learn these compliance probabilities efficiently. Twin-MinKPDR-CB works for non-monotone reward functions, bounded minknapsack constraints, and time-varying optimal sets. We find that Twin-MinKPDR-CB achieves sub-linear regret of O(log T) with T being the number of rounds demand response is run.

Abstract

Generative Adversarial Networks (GANs) are by far the most successful generative models. Learning the transformation which maps a low dimensional input noise to the data distribution forms the foundation for GANs. Although they have been applied in various domains, they are prone to certain challenges like mode collapse and unstable training. To overcome the challenges, researchers have proposed novel loss functions, architectures, and optimization methods. In our work here, unlike the previous approaches, we focus on the input noise and its role in the generation. We aim to quantitatively and qualitatively study the effect of the dimension of the input noise on the performance of GANs. For quantitative measures, typically Fr´echet Inception Distance (FID) and Inception Score (IS) are used as performance measure on image data-sets. We compare the FID and IS values for DCGAN and WGAN-GP. We use three different image data-sets – each consisting of different levels of complexity. Through our experiments, we show that the right dimension of input noise for optimal results depends on the data-set and architecture used. We also observe that the state of the art performance measures does not provide enough useful insights. Hence we conclude that we need further theoretical analysis for understanding the relationship between the low dimensional distribution and the generated images. We also require better performance measures.

Abstract

Bitcoin is the first of its kind, a truly decentralized and anonymous cryptocurrency. To realize it, it has developed a blockchain technology using the concept of ‘Proof of Work’ (PoW). The miners, nodes responsible for writing transaction database, solve a cryptographic puzzle to claim the right to write to the database. Though bitcoin and many other relevant cryptocurrencies such as ether use revolutionary ideas, the main criticism involves the computing resource and energy consump- tion to solve the puzzles that have otherwise no use. There are attempts to use the PoW to do something useful, commonly referred to as Proof-of-Useful-Work (PoUW). In this paper, we attempt to (i) make PoUW more usable – describe how a central problem setter can crowdsource their work as PoUW and (ii) in the true spirit of blockchains, decentralize the role of problem setter, whom we call puzzlers. We propose a formal framework to do so, namely PUPoW. PUPoW has an inbuilt provision of payments from puzzler to the miner who solves its puzzle. Additionally, miners have the option to not rely on continuous feed of the puzzles and instead use original PoW puzzles. We also propose a way to use PUPoW for solving TOR vanity URL generation and bitcoin vanity address generation problems. We call this PUPoW blockchain solving vanity address generation problems as VanityCoin. Both the problems need to generate public keys from private keys such that resultant addresses are of interest. Such key pairs are found only by a brute force search. However, there are privacy concerns that miners would know the private keys of the puzzlers. We resolve this by splitting the private keys, and the miners would know only one part of it. In summary, we are proposing how PoW can be made practically useful, and we believe such an approach is needed for PoW blockchains to survive.

Abstract

Several optimization scenarios involve multiple agents that desire to protect the privacy of their preferences. There are distributed algorithms for constraint optimization that provide improved privacy protection through secure multiparty computation. However, it comes at the expense of high computational complexity and does not constitute a rigorous privacy guarantee for optimization outcomes, as the result of the computation itself may compromise agents’ preferences. In this work, we show how to achieve privacy, specifically differential privacy, through the randomization of the solving process. In particular, we present P-Gibbs, which adapts the SD-Gibbs algorithm to obtain differential privacy guarantees with much higher computational efficiency. Experiments on graph coloring and meeting scheduling show the algorithm’s privacy-performance trade-off for varying privacy budgets, and the SD-Gibbs algorithm.

Abstract

We introduce MeronymNet, a novel hierarchical approach for con- trollable, part-based generation of multi-category objects using a single unified model. We adopt a guided coarse-to-fine strategy involving semantically conditioned generation of bounding box layouts, pixel-level part layouts and ultimately, the object depic- tions themselves. We use Graph Convolutional Networks, Deep Recurrent Networks along with custom-designed Conditional Vari- ational Autoencoders to enable flexible, diverse and category-aware generation of 2-D objects in a controlled manner. The performance scores for generated objects reflect MeronymNet’s superior perfor- mance compared to multiple strong baselines and ablative variants.

Abstract

The current study aims to analyse VR measures in correspondence with 360° virtual affective experiences to understand the underlying spatio-temporal attention processes. We selected extreme pleasant and extreme unpleasant emotional videos from the Stanford affective database and displayed to participants on HMD VR. The participants’ task was to explore the video and report their affective experience using the Self- Assessment Manikin (SAM) scale. The exploration behaviour was analysed using head-tracking parameters, such as standard deviation of all the three head-rotation axes (yaw, pitch and roll), angular speed and region-wise analysis. We observed a positive correlation between standard deviation of yaw and valence, and angular speed and valence. The result suggests that affective experience not only increases the scope of attention spatially (wide scanning), it also motivates to seek more information that leads to more head-movement during exploration. In region- wise analysis, we observed front field-of-view was primarily explored, and rear field-of-view was least explored, showing natural responses similar to real environments. In addition, we used Beck Depression Inventory-II (BDI) to assess participants’ emotional state, specifically depression symptoms. The relation between virtual affective experience and 360° exploration be- haviour is discussed in light of the BDI categories. The study findings broaden the scope for VR affective research by enabling psychomotor assessment in more ecologically valid settings. Index Terms—Emotion, Depression, Affective State, 360° Ex- ploration, Head-Tracking

Abstract

Control systems operate industrial plants to accomplish stakeholder objectives like achieving production targets, complying with environmental ordinances, handling faults, etc. Such stakeholder objectives get realised by identifying and executing valid control actions on the plant’s control system. E.g., to achieve fault management a command is fired to place machines in a fault mode when the plant is under an error state. Arriving at such control actions is a non-trivial task demanding a detailed understanding of the plant’s structure and behaviour. Besides, it is also essential to verify the consequences of such control actions relative to other cross-cutting objectives and plant behaviour. E.g., to fulfil fault management objectives, the action to set machines in fault mode may affect production goals due to the machine unavailability. Hence, validation of control actions is vital before executing them using the actual plant’s control system. With digital twin technologies (DT), it is now possible to verify the implications of such control actions against a plant’s behaviour and objectives in a simulated environment without affecting the actual plant operations. DTs get developed autonomously as one-off solutions to simulate and validate plant control actions in the current state of practice, demanding high efforts and domain expertise. Our paper proposes a knowledge-driven approach enabling automation in DT development. The result of our approach is an auto-generated digital twin that pro-actively mimics the plant’s control system behaviour and helps with the validation of control actions before their execution. We use this approach to build three fault management DTs in a power plant. The application of our approach significantly reduces the manual efforts and development time to build such DTs.

Abstract

This Full Paper in the Innovative Practice category begins by asking “can algorithms be thought of and taught as dynamical systems?” Our exploration of this idea — Algodynamics — is guided by a vision to achieve convergence between computing and engineering education. The engineering sciences share a common conceptual vocabulary originating in dynamical systems: state spaces, flows, actions, invariants, fixed points, convergence, etc. The goal of algodynamics is to build, ab initio, a framework for understanding and teaching algorithms using concepts from dynamics. This allows us to teach computing and algorithms as an engineering science. Engineers work with models. In algodynamics, models are expressed using transition systems rather than as pseudo- code or programs. This allows a crisp representation of two important classes of computation: sequential algo- rithms as ‘discrete flows’ (iterative systems) and interactive applications as ‘action flows’ (transition systems). The focus of this paper is on the first of these classes, algorithms, but using ideas from the second, viz., tran- sition systems. In this framework, algorithms emerge as convergent iterative systems. Due to their non-interactivity, iterative systems may be hard to understand. The student may trace through the algorithm, but to know how the parts of the algorithm work together requires ‘opening up’ the algorithm and situating it within the more general class of interactive systems. Doing so helps the student to understand the machinery of an algorithm in an incremen- tal and modular way. The student interactively solves the algorithmic problem, experimenting with various strategies along the way. At each stage of the design, interactivity is traded for automation

Abstract

Precise boundary annotations of image regions can be crucial for downstream applications which rely on region-class semantics. Some document collections contain densely laid out, highly irregular and over- lapping multi-class region instances with large range in aspect ratio. Fully automatic boundary estimation approaches tend to be data intensive, cannot handle variable-sized images and produce sub-optimal results for aforementioned images. To address these issues, we propose Bound- aryNet, a novel resizing-free approach for high-precision semi-automatic layout annotation. The variable-sized user selected region of interest is first processed by an attention-guided skip network. The network opti- mization is guided via Fast Marching distance maps to obtain a good quality initial boundary estimate and an associated feature represen- tation. These outputs are processed by a Residual Graph Convolution Network optimized using Hausdorff loss to obtain the final region bound- ary. Results on a challenging image manuscript dataset demonstrate that BoundaryNet outperforms strong baselines and produces high-quality se- mantic region boundaries. Qualitatively, our approach generalizes across multiple document image datasets containing different script systems and layouts, all without additional fine-tuning. We integrate Bound- aryNet into a document annotation system and show that it provides high annotation throughput compared to manual and fully automatic alternatives. Keywords: document layout analysis · interactive · deep learning