This paper is a contribution to neural network semantics, a foundational framework for neuro-symbolic AI. The key insight of this theory is that logical operators can be mapped to operators on neural network states. In this paper, we do this for a neural network learning operator. We map a dynamic operator [φ] to iterated Hebbian learning, a simple learning policy that updates a neural network by repeatedly applying Hebb's learning rule until the net reaches a fixed-point. Our main result is that we can "translate away" [φ]-formulas via reduction axioms. This means that completeness for the logic of iterated Hebbian learning follows from completeness of the base logic. These reduction axioms also provide (1) a human-interpretable description of iterated Hebbian learning as a kind of plausibility upgrade, and (2) an approach to building neural networks with guarantees on what they can learn.

The generalized symmetric group, or the group of colored permutations, is defined by the wreath product <?TeX $S(m,n)=\mathbb {Z}_m\wr S_n$?> Math 1 , where <?TeX $\mathbb {Z}_m$?> Math 2 denotes the cyclic group of order m and Sn denotes the symmetric group of degree n. We extend the notion of permutation classes, downward closed sets under containment, to colored permutations. The enumeration and growth rate of permutation classes have long been studied. For instance, if <?TeX $\mathcal {C}$?> Math 3 denotes a permutation class, then <?TeX $|\mathcal {C}\cap S_n|< F_n$?> Math 4 for some n, if and only if <?TeX $|\mathcal {C}\cap S_n|$?> Math 5 is eventually polynomial. This result is known as the Fibonacci dichotomy. We prove that the Fibonacci dichotomy also holds for colored permutation classes and extend the known algorithm that produces the polynomial in the case of permutations to colored permutations. We implement the algorithm and include applications to classes of colored permutations that require a fixed number of prefix reversals or block reversals to be sorted.

In this paper, we present an algorithm that enumerates a certain class of signed permutations, referred to as grid signed permutation classes. In the case of permutations, the corresponding grid classes are of interest because they are equivalent to the permutation classes that can be enumerated by polynomials. Furthermore, we apply our results to genome rearrangements and establish that the number of signed permutations with fixed prefix reversal and reversal distance is given by polynomials that can be computed by our algorithm.

The use of automatic grading tools has become nearly ubiquitous in large undergraduate programming courses, and recent work has focused on improving the quality of automatically generated feedback. However, there is a relative lack of data directly comparing student outcomes when receiving computer-generated feedback and human-written feedback. This paper addresses this gap by splitting one 90-student class into two feedback groups and analyzing differences in the two cohorts' performance. The class is an intro to AI with programming HW assignments. One group of students received detailed computer-generated feedback on their programming assignments describing which parts of the algorithms' logic was missing; the other group additionally received human-written feedback describing how their programs' syntax relates to issues with their logic, and qualitative (style) recommendations for improving their code. Results on quizzes and exam questions suggest that human feedback helps students obtain a better conceptual understanding, but analyses found no difference between the groups' ability to collaborate on the final project. The course grade distribution revealed that students who received human-written feedback performed better overall; this effect was the most pronounced in the middle two quartiles of each group. These results suggest that feedback about the syntax-logic relation may be a primary mechanism by which human feedback improves student outcomes.

POGIL activities have been used for various computer science courses. However, there is no published POGIL activity for introducing parallel and distributed computing concepts. Recent ABET curriculum recommendations include the introduction of parallel and distributed computing concepts in undergraduate Computer Science/Engineering programs. In this work, we plan a cross-university study of evaluating the impact of using POGIL to introduce parallel computing topics in Data Structures and Algorithms or similar courses in an undergraduate computer science curriculum. We designed a POGIL tool that includes an unplugged activity that helps demonstrate some fundamental concepts of parallel computing. This unplugged activity is then followed by a set of reflective questions regarding potential advantages and challenges of using parallel computing. %NEEDED? WE SAY IT LATER: The POGIL tool also contains code written in Java, C/C++, and C\#. We plan to use PRE/POST surveys to collect data from undergraduate CS students from five universities located in different parts of the US with diverse student population. At the end of this work in progress, we will use the data to investigate how this POGIL activity helps students gain an understanding of parallel computing.

We consider the thresholding bandit problem, whose goal is to find arms of mean rewards above a given threshold $\theta$, with a fixed budget of $T$ trials. We introduce LSA, a new, simple and anytime algorithm that aims to minimize the aggregate regret (or the expected number of mis-classified arms). We prove that our algorithm is instance-wise asymptotically optimal. We also provide comprehensive empirical results to demonstrate the algorithm's superior performance over existing algorithms under a variety of different scenarios.

We study the best arm identification problem in linear bandits, where the mean reward of each arm depends linearly on an unknown $d$-dimensional parameter vector $\theta$, and the goal is to identify the arm with the largest expected reward. We first design and analyze a novel randomized $\theta$ estimator based on the solution to the convex relaxation of an optimal $G$-allocation experiment design problem. Using this estimator, we describe an algorithm whose sample complexity depends linearly on the dimension $d$, as well as an algorithm with sample complexity dependent on the reward gaps of the best $d$ arms, matching the lower bound arising from the ordinary top-arm identification problem. We finally compare the empirical performance of our algorithms with other state-of-the-art algorithms in terms of both sample complexity and computational time.

In this paper, we describe the active learning and collaborative learning activities implemented in an introductory mid-size discrete mathematics course for Informatics majors. Active learning and collaborative learning have been used to increase student engagement, but incorporating them in smaller classes is a completely different experience from doing so in larger classes. We offer some tips and suggestions on how to incorporate these activities in larger classes including the utilization of undergraduate teaching assistants during lectures, and allowing students to work together on worksheets during lectures with the help of the teaching staff. Course questionnaires collected from five different sections that ran in the spring, summer, and fall of 2016 with around 60 to 70 students suggest that this approach has been well-received. Furthermore, the DFW rate (the proportion of students that received a D, F, or withdrew from the class) of these sections was lower than the DFW rate of other sections that shared the same evaluations (exams, homework assignments, and quizzes) and grading scheme to determine the final letter grade.

Accurate, efficient, global observation of natural events is important for ecologists, meteorologists, governments, and the public. Satellites are effective but limited by their perspective and by atmospheric conditions. Public images on photo-sharing websites could provide crowd-sourced ground data to complement satellites, since photos contain evidence of the state of the natural world. In this work, we test the ability of computer vision to observe natural events in millions of geo-tagged Flickr photos, over nine years and an entire continent. We use satellites as (noisy) ground truth to train two types of classifiers, one that estimates if a Flickr photo has evidence of an event, and one that aggregates these estimates to produce an observation for given times and places. We present a web tool for visualizing the satellite and photo observations, allowing scientists to explore this novel combination of data sources.

As educators want to try to find new ways to engage our students in and out of the classroom, allowing them to enhance their learning as well as their overall experience of the class. Active learning techniques challenge students to learn in a "non-traditional" way by developing critical thinking skills and having a little fun at the same time. While our students become more effective students, we can also become more effective educators, as we can quickly and accurately assess learning outcomes.