Artificial Intelligence and Robotics Commons^™

What-Is And How-To For Fairness In Machine Learning: A Survey, Reflection, And Perspective, Zeyu Tang, Jiji Zhang, Kun Zhang

Machine Learning Faculty Publications

Algorithmic fairness has attracted increasing attention in the machine learning community. Various definitions are proposed in the literature, but the differences and connections among them are not clearly addressed. In this paper, we review and reflect on various fairness notions previously proposed in machine learning literature, and make an attempt to draw connections to arguments in moral and political philosophy, especially theories of justice. We also consider fairness inquiries from a dynamic perspective, and further consider the long-term impact that is induced by current prediction and decision. In light of the differences in the characterized fairness, we present a flowchart …

Learning To Control Under Time-Varying Environment, Yuzhen Han, Ruben Solozabal, Jing Dong, Xingyu Zhou, Martin Takac, Bin Gu

Machine Learning Faculty Publications

This paper investigates the problem of regret minimization in linear time-varying (LTV) dynamical systems. Due to the simultaneous presence of uncertainty and non-stationarity, designing online control algorithms for unknown LTV systems remains a challenging task. At a cost of NP-hard offline planning, prior works have introduced online convex optimization algorithms, although they suffer from nonparametric rate of regret. In this paper, we propose the first computationally tractable online algorithm with regret guarantees that avoids offline planning over the state linear feedback policies. Our algorithm is based on the optimism in the face of uncertainty (OFU) principle in which we optimistically …

Flecs: A Federated Learning Second-Order Framework Via Compression And Sketching, Artem Agafonov, Dmitry Kamzolov, Rachael Tappenden, Alexander Gasnikov, Martin Takac

Machine Learning Faculty Publications

Inspired by the recent work FedNL (Safaryan et al, FedNL: Making Newton-Type Methods Applicable to Federated Learning), we propose a new communication efficient second-order framework for Federated learning, namely FLECS. The proposed method reduces the high-memory requirements of FedNL by the usage of an L-SR1 type update for the Hessian approximation which is stored on the central server. A low dimensional 'sketch' of the Hessian is all that is needed by each device to generate an update, so that memory costs as well as number of Hessian-vector products for the agent are low. Biased and unbiased compressions are utilized to …

Offline Reinforcement Learning With Causal Structured World Models, Zheng-Mao Zhu, Xiong-Hui Chen, Hong-Long Tian, Kun Zhang, Yang Yu

Machine Learning Faculty Publications

Model-based methods have recently shown promising for offline reinforcement learning (RL), aiming to learn good policies from historical data without interacting with the environment. Previous model-based offline RL methods learn fully connected nets as world-models to map the states and actions to the next-step states. However, it is sensible that a world-model should adhere to the underlying causal effect such that it will support learning an effective policy generalizing well in unseen states. In this paper, We first provide theoretical results that causal world-models can outperform plain world-models for offline RL by incorporating the causal structure into the generalization error …

What’S So Artificial And Intelligent About Artificial Intelligence? A Conceptual Framework For Ai, Rebekah L. H. Rice

SPU Works

There is currently a good deal of attention being focused on artificial intelligence, broadly speaking, and deep learning, specifically. The attention is warranted, as these technologies are predicted to affect our collective lives in innumerable ways even beyond their already expansive social reach. There is much to consider regarding the benefits and potential harms of AI. And of course there are the apocalyptic musings about super-intelligent machines running amok, bringing science fiction scenarios uncomfortably close to anticipated reality. But productively engaging in discussions about the ethical and social implications of AI, and about which sorts of futures it is reasonable …

A Theological Framework For Reflection On Artificial Intelligence, Michael D. Langford

SPU Works

The theological questions before us in a digital age are pressing. What does God think of AI? Is AI good or evil? Will AI save us? What sort of future will AI give us? In what follows, I want to briefly introduce a few theological concepts that will hopefully help equip us for theological reflection on AI. We will begin with the question of epistemology, or how it is that we come by knowledge; in the realm of theology, this centers on revelation. We will then touch on the doctrine of creation, including the understanding of what it means to …

Artificial Intelligence And Theological Personhood, Michael D. Langford

SPU Works

Can AI be a person? What does God tell us about humanity and personhood? These are questions of theological anthropology and involve inquiring after the nature of humanity as God’s creation and what God wills for human personhood.

To address these inquiries, we will look at three biblical texts that bear on issues of theological anthropology, hopefully garnering some theological resources to consider the anthropological status of AI. Specifically, we will look at three “creation” texts that necessarily deal with the nature of human personhood within the divine economy of salvation history. The first is Genesis 1 and 2, which …

Reinforcement In The Information Revolution, Phillip M. Baker

SPU Works

This chapter will outline what it means to be a behaving human and how AI makes sense of these concepts. It will then explore possible near-future implications of our remarkable progress in understanding how human behavior works with the assistance of AI from a neurobiological basis. A focus on understanding the reinforcement mechanisms of the brain will reveal the consequences of ceding control of so much of our brain-environment interactions to AI. It will conclude by offering a potential Christian response to this digital reality from a uniquely Anabaptist perspective.

An Introduction To Artificial Intelligence, Carlos R. Arias

SPU Works

This chapter explores the evolution of artificial intelligence, starting with the first ideas of Alan Turing, going through the promises of its inception, and landing in our current state, when AI invokes a sense of power and awe. Next, the chapter will provide a summary of different technologies related to AI and machine learning, such as deep neural networks, to help the reader distinguish different terminologies. The chapter will end with a discussion of some potential tendencies concerning how AI may be used or evolve in the near future, and some questions about the technology in the long term.

Sin And Grace, Bruce D. Baker

SPU Works

The theological lens of sin and grace gives a broader and deeper viewpoint than mere ethics. Ethical analysis is of course useful and necessary, but ethics alone is not enough. Ethics apart from a robust, holistic understanding of humans as persons-in-communion will remain mired in reductionist thinking about human dignity and morality. Therefore, this final chapter addresses the ethical issues of AI through the lens of sin and grace.

Epilogue: A Litany For Faithful Engagement With Artificial Intelligence, Bruce D. Baker

SPU Works

A litany is a thoughtfully organized prayer for use in public worship by the church, or as a personal devotional practice by individuals. This seems a fitting way to close our reflection on AI, faith, and the future. Prayer will be essential to our faithful response to the new opportunities and challenges AI brings. Our hope is that this litany will serve as a practical guide to thoughtful invocation of the Holy Spirit in prayers for wisdom and discernment, and in the daily disciplines of spiritual growth.

21st Century Learning Skills And Artificial Intelligence, David Wicks, Michael Paulus

SPU Works

The chapter explores four concepts important for learning and AI in the twenty-first century—creativity, critical thinking, communication, and collaboration (the “4Cs”)—as well as reflections on the theological significance of creativity and community.

Automation And Apocalypse: Imagining The Future Of Work, Michael Paulus

SPU Works

This chapter provides an orientation to the history of technology, work, and the theology of work and then explores three visions of the future of work—a literary dystopia, a philosophical utopia, and a theological apocalypse—as resources for understanding the significance of work and imagining its future. In the first vision, found in Kurt Vonnegut’s speculative novel Player Piano, automation leads to the end of meaningful work and nearly renders humans obsolete. This dystopic vision reveals the value of human work but remains skeptical about our ability to preserve it against the advances of automation. The second vision comes from …

Introduction, Michael Paulus

SPU Works

Artificial intelligence is rapidly and radically changing our lives and world. This book is a multidisciplinary engagement with the present and future impacts of AI from the standpoint of Christian faith. It provides technological, philosophical, and theological foundations for thinking about AI, as well as a series of reflections on the impact of AI on relationships, behavior, education, work, and moral action. The book serves as an accessible introduction to AI as well as a guide to wise consideration, design, and use of AI by examining foundational understandings and beliefs from a Christian perspective.

Training Thinner And Deeper Neural Networks: Jumpstart Regularization, Carles Riera, Camilo Rey, Thiago Serra, Eloi Puertas, Oriol Pujol

Faculty Conference Papers and Presentations

Neural networks are more expressive when they have multiple layers. In turn, conventional training methods are only successful if the depth does not lead to numerical issues such as exploding or vanishing gradients, which occur less frequently when the layers are sufficiently wide. However, increasing width to attain greater depth entails the use of heavier computational resources and leads to overparameterized models. These subsequent issues have been partially addressed by model compression methods such as quantization and pruning, some of which relying on normalization-based regularization of the loss function to make the effect of most parameters negligible. In this work, …

Monofacial Vs Bifacial Solar Photovoltaic Systems In Snowy Environments, Koami Soulemane Hayibo, Aliaksei Petsiuk, Pierce Mayville, Laura Brown, Joshua M. Pearce

Electrical and Computer Engineering Publications

There has been a recent surge in interest in the more accurate snow loss estimates for solar photovoltaic (PV) systems as large-scale deployments move into northern latitudes. Preliminary results show bifacial modules may clear snow faster than monofacial PV. This study analyzes snow losses on these two types of systems using empirical hourly data including energy, solar irradiation and albedo, and open-source image processing methods from images of the arrays in a northern environment in the winter. Projection transformations based on reference anchor points and snowless ground truth images provide reliable masking and optical distortion correction with fixed surveillance cameras. …

Multiple Object Tracking For Marine Science, Nicholas A. Wachter

Computer Science and Software Engineering

As drone and computer vision technology has been improving, many fields of study have been quick to utilize it to improve the accuracy and ease of data collection. The combination of the two technologies is perfect for surveying large areas and identifying features of interest. Marine science utilizes these technologies for activities such as animal tracking and population counting. I am working with the Drones for Marine Science research group at Cal Poly who want to build a fleet of drones that will fly out over the ocean to identify and track various marine animals. My role will be to …

Reinforcement Learning-Based Interactive Video Search, Zhixin Ma, Jiaxin Wu, Zhijian Hou, Chong-Wah Ngo

Research Collection School Of Computing and Information Systems

Despite the rapid progress in text-to-video search due to the advancement of cross-modal representation learning, the existing techniques still fall short in helping users to rapidly identify the search targets. Particularly, in the situation that a system suggests a long list of similar candidates, the user needs to painstakingly inspect every search result. The experience is frustrated with repeated watching of similar clips, and more frustratingly, the search targets may be overlooked due to mental tiredness. This paper explores reinforcement learning-based (RL) searching to relieve the user from the burden of brute force inspection. Specifically, the system maintains a graph …

Group Contextualization For Video Recognition, Yanbin Hao, Hao Zhang, Chong-Wah Ngo, Xiangnan He

Research Collection School Of Computing and Information Systems

Learning discriminative representation from the complex spatio-temporal dynamic space is essential for video recognition. On top of those stylized spatio-temporal computational units, further refining the learnt feature with axial contexts is demonstrated to be promising in achieving this goal. However, previous works generally focus on utilizing a single kind of contexts to calibrate entire feature channels and could hardly apply to deal with diverse video activities. The problem can be tackled by using pair-wise spatio-temporal attentions to recompute feature response with cross-axis contexts at the expense of heavy computations. In this paper, we propose an efficient feature refinement method that …

Comparing Learned Representations Between Unpruned And Pruned Deep Convolutional Neural Networks, Parker Mitchell

Master's Theses

While deep neural networks have shown impressive performance in computer vision tasks, natural language processing, and other domains, the sizes and inference times of these models can often prevent them from being used on resource-constrained systems. Furthermore, as these networks grow larger in size and complexity, it can become even harder to understand the learned representations of the input data that these networks form through training. These issues of growing network size, increasing complexity and runtime, and ambiguity in the understanding of internal representations serve as guiding points for this work.

In this thesis, we create a neural network that …