This innovative practice full paper presents a guided approach to integrating universal design for learning (UDL) principles into an Explicit Instruction classroom. To address the common challenges of scope and additional time requirements of integrating UDL, we are using generative artificial intelligence (GAI) tools like ChatGPT because they have reached a level of functionality that allows a GAI tool to replace multiple specialized tools. This paper provides specific guidelines of where UDL interventions can be included in an explicit instruction lesson and how to use GAI to support them. This paper has multiple goals. First, we are focused on an approach that can help embed modern, best practices in traditional, higher ed STEM classes that use an approach that is typically close to the explicit instruction model. Second, we want to show how GAI can be used to implement UDL principles in classes in a way that does not require much additional instructor time and can help UDL adoption scale to larger classes. Example GAI prompts and “real” responses are provided for those who are unfamiliar with GAI tools, in order to help demonstrate the capabilities of the tools.

This research-to-practice full paper presents and approach to bringing convergence to the undergraduate engineering context. Convergence is the process of integrating a variety of ideas, skills, and methods to create new ideas, skills, and methods in order to address complex, socially relevant challenges like the UN Sustainable Development Goals [1] and the National Academy of Engineering's (NAE) Grand Challenges [2]. In the US, the National Science Foundation (NSF) has been a major driver of convergence related research and has focused on work primarily at the graduate level and beyond. To explore how convergence concepts translate to an undergraduate engineering context this research to practice paper describes a taxonomy that translates convergent knowledge, skills, and mindsets into the domain of undergraduate engineering education. While we do not believe it is reasonable to expect undergraduates to engage with convergence in the same way as graduate students or postdoctoral scholars, we believe that they can develop in areas that will allow them to engage in convergent work later in their careers. This paper first defines convergence and then examines the challenges and opportunities related to developing a student's ability to do convergent work in an undergraduate context. The developed taxonomy outlines the knowledge, skills, mindsets, and structures that support convergent work from the larger research literature, and adapts these to an undergraduate context. The taxonomy is then used to conduct a gap analysis of an undergraduate electrical and computer engineering degree program. This analysis is based on the syllabi. This work was conducted in the context of an electrical and computer engineering department situated in a medium-sized primarily undergraduate liberal arts institution in the mid-Atlantic region. As the challenges and opportunities are similar to but also unique to this institution this work forms a rich case study that can inform similar efforts in other institutions and contexts where a similar gap analysis may be beneficial. The goal of this work is to enable others to analyze an their existing student experience to see what aspects of convergence are currently included.

For internet users it is frequently challenging to properly navigate and interpret the overwhelming amount of information available to them. Multimodal artificial intelligence (AI) systems can aid in the process; for example, by helping to debunk misinformation, or by placing video content into a proper context. The training of commensurate AI systems requires the labelling of sizable multimodal datasets, a process that is very time-consuming and prone to being outpaced by the rapid changes of the information landscape. To address this problem, we are introducing a hybrid fusion algorithm, Bimodal Gaussian-Logits glObal stream Weighting (BiGLOW), which performs effective information integration in scenarios in which only a very limited amount of bimodally labelled training data is available. The method is motivated from Bayesian inference; we model a modality-weighted linear combination of differences in logits between class pairs through a Gaussian distribution. We evaluated the proposed method with two bimodal multiclass datasets: the Fakeddit dataset for fake news classification and the TAU Audio-Visual Urban Scenes 2021 for urban scene identification. The two datasets capture four of the most frequently used modalities on the web: text, images, audio, and video. Scenarios with scarce training data were created by reducing the respective size of the bimodally labelled set to various small fractions of the overall data size. Our experimental results indicate that the proposed hybrid fusion algorithm surpasses competing purely neural-network-based models in terms of the model complexity, average accuracy, robustness, and time requirements for both model training and evaluation. We attribute the improved performance of the proposed method to its implied, Bayesian inference inspired regularization.

Education literature has long emphasized the compounding benefits of reflective practice. Although reflection has largely been used as a tool for developing writing skills, contem-porary research has explored its contributions to other disciplines including professional occupations such as nursing, teaching and engineering. Reflective assignments encourage engineering students to think critically about the impact engineers can and should have in the global community and their future role in engineering. The Department of Electrical and Computer Engineering at a small liberal arts college adopted ePortfolios in a first-year design course to encourage students to reframe their experiences and cultivate their identities as engineers. Our recent work demonstrated that students who create ePortfolios cultivate habits of reflective thinking that continue in subsequent courses within our program's design sequence. However, student ability to transfer reflective habits across domains has remained unclear and encouraging critical engagement beyond the focused scope of technical content within more traditional core engineering courses is often difficult. In this work, we analyze students' ability to transfer habits of reflective thinking across domains from courses within a design-focused course sequence to technical content-focused courses within a degree program. Extending reflection into core courses in a curriculum is important for several reasons. First, it stimulates metacognition which enables students to transfer content to future courses. Second, it builds students' ability to think critically about technical subject matter. And third, it contributes to the ongoing development of their identities as engineers. Particularly for students traditionally underrepresented in engineering, the ability to integrate prior experiences and interests into one's evolving engineering identity may lead to better retention and sense of belonging in the profession. In the first-year design course, electrical and computer engi-neering students (N=28) at a liberal arts university completed an ePortfolio assignment to explore the discipline. Using a combination of inductive and deductive coding techniques, mul-tiple members of our team coded student reports and checked for intercoder reliability. Previously, we found that students' reflection dramatically improved in the second-year design course [1]. Drawing upon Hatton and Smith's (1995) categorizations of reflective thinking [2], we observed that students were particu-larly proficient in Dialogic Reflection, or reflection that relates to their own histories, interests, and experiences. In this paper, we compare the quality of student reflections in the second-year design course with those in a second-year required technical course to discover if reflective capabilities have transferred into a technical domain. We discovered that students are able to transfer reflective thinking across different types of courses, including those em-phasizing technical content, after a single ePortfolio activity. Furthermore, we identified a similar pattern of improvement most notably in Dialogic Reflection. This finding indicates that students are developing sustained habits of reflective thinking. As a result, we anticipate an increase in their ability to retain core engineering concepts throughout the curriculum. Our future plans are to expand ePortfolio usage to all design courses as well as some fundamental technical courses throughout the curriculum.

Wicked problems, the National Academy of Engineering's Grand Challenges, the United Nations' Sustainability Goals, and similar complex, global-scale endeavors fall under the broad umbrella of “convergent” work. Over the past two decades there has been an increase in interest and funding for work in this space. The NSF has two programs focused in this area, Growing Convergence Research and the Convergence Accelerator. Boston University's College of Engineering recently announced a focus on convergent projects and work. The National Academic of Engineering also has the Grand Challenge Scholars program with over 100 participating schools. The list continues to grow. The broad concept of convergence seems to be quite simple: combine the ideas, skills, and/or methods of multiple disciplines to create something new. More specific definitions vary and while the interest in convergence and convergent problems continues to increase, there is no easily operational definition of convergence. This is especially true with respect to undergraduate-level education where students have limited experience and knowledge to carry out such efforts. To better understand the variation that exists within the literature on convergence we conducted a systematic review to explore how convergence is defined in scholarly literature. We have identified a small number of categories within the definition space and conducted a thematic analysis of the aspects of each. The results show that there is a fairly consistent focus on the work being socially-relevant and on creating something new such as an idea, method, product, or process to address desired needs. Additionally, doing convergent work requires the integration of aspects of multiple disciplines and is conducted by diverse teams. Lastly, the disciplinary backgrounds of those teams almost always includes the natural and biological sciences with a subset the following disciplines: information or computing sciences, engineering, social sciences, and humanities. While there is some consistency in the definition, there also seems to be space for some variation which leaves for some level of choice in the definition.

This work in progress paper addresses one of the future challenges facing academic disciplines-traditional STEM as well as the social sciences and humanities-of how to prepare students to address complex problems that require a range of disciplinary perspectives. The goal of such preparation is termed convergence. Convergence captures both how different sets of expertise become focused in solving a problem as well as the network of connections that is built in undertaking these activities. Efforts at convergence are usually focused at practitioner, postgraduate, and graduate levels and engage multiple disciplines. Here the authors report on early-phase development of an effort to integrate convergence education into an undergraduate disciplinary-based degree program in electrical and computer engineering focused on integrating problems and projects across the curriculum. Key challenges faced were how to distinguish engineering topics from societal concerns in ways that were meaningful for students, balancing discipline-specific skills with fostering systemic understanding, and identifying relevant projects with attainable goals. Existing frameworks serve as a foundation for students to understand systemic and convergent issues. Framework development by individual students is scaffolded by expanding grading practices to provide feedback on skills believed to support convergence, developing ways to elicit student narratives about how courses and projects relate to individual interests, and adopting learning technologies that can support more emphasis on projects throughout the curriculum. Initial results on changing grading practices and introducing projects that are grounded outside of disciplinary context are presented. The relevance of this work to engineering education arises from the relatively small amount of empirical work exploring how to prepare engineering undergraduates to address convergent problems as well as the importance of such problems. The innovative potential of the work is that the effort will eventually become curriculum-wide and supported by structural changes to grading practices.

Traditional computational authorship attribution describes a classification task in a closed-set scenario.Given a finite set of candidate authors and corresponding labeled texts, the objective is to determine which of the authors has written another set of anonymous or disputed texts.In this work, we propose a probabilistic autoencoding framework to deal with this supervised classification task.Variational autoencoders (VAEs) have had tremendous success in learning latent representations.However, existing VAEs are currently still bound by limitations imposed by the assumed Gaussianity of the underlying probability distributions in the latent space.In this work, we are extending a VAE with an embedded Gaussian mixture model to a Student-t mixture model, which allows for an independent control of the "heaviness" of the respective tails of the implied probability densities.Experiments over an Amazon review dataset indicate superior performance of the proposed method.

Authorship verification is the task of analyzing the linguistic patterns of two or more texts to determine whether they were written by the same author or not. The analysis is traditionally performed by experts who consider linguistic features, which include spelling mistakes, grammatical inconsistencies, and stylistics for example. Machine learning algorithms, on the other hand, can be trained to accomplish the same, but have traditionally relied on so-called stylometric features. The disadvantage of such features is that their reliability is greatly diminished for short and topically varied social media texts. In this interdisciplinary work, we propose a substantial extension of a recently published hierarchical Siamese neural network approach, with which it is feasible to learn neural features and to visualize the decision-making process. For this purpose, a new large-scale corpus of short Amazon reviews for text comparison research is compiled and we show that the Siamese network topologies outperform state-of-the-art approaches that were built up on stylometric features. Our linguistic analysis of the internal attention weights of the network shows that the proposed method is indeed able to latch on to some traditional linguistic categories.

Authorship verification tries to answer the question if two documents with unknown authors were written by the same author or not. A range of successful technical approaches has been proposed for this task, many of which are based on traditional linguistic features such as n-grams. These algorithms achieve good results for certain types of written documents like books and novels. Forensic authorship verification for social media, however, is a much more challenging task since messages tend to be relatively short, with a large variety of different genres and topics. At this point, traditional methods based on features like n-grams have had limited success. In this work, we propose a new neural network topology for similarity learning that significantly improves the performance on the author verification task with such challenging data sets.

Audio-visual speech recognition is a promising approach to tackling the problem of reduced recognition rates under adverse acoustic conditions. However, finding an optimal mechanism for combining multi-modal information remains a challenging task. Various methods are applicable for integrating acoustic and visual information in Gaussian-mixture-model-based speech recognition, e.g., via dynamic stream weighting. The recent advances of deep neural network (DNN)-based speech recognition promise improved performance when using audio-visual information. However, the question of how to optimally integrate acoustic and visual information remains. In this paper, we propose a state-based integration scheme that uses dynamic stream weights in DNN-based audio-visual speech recognition. The dynamic weights are obtained from a time-variant reliability estimate that is derived from the audio signal. We show that this state-based integration is superior to early integration of multi-modal features, even if early integration also includes the proposed reliability estimate. Furthermore, the proposed adaptive mechanism is able to outperform a fixed weighting approach that exploits oracle knowledge of the true signal-to-noise ratio.

We present a new approach for corpus-based speech enhancement that significantly improves over a method published by Xiao and Nickel in 2010. Corpus-based enhancement systems do not merely filter an incoming noisy signal, but resynthesize its speech content via an inventory of pre-recorded clean signals. The goal of the procedure is to perceptually improve the sound of speech signals in background noise. The proposed new method modifies Xiao's method in four significant ways. Firstly, it employs a Gaussian mixture model (GMM) instead of a vector quantizer in the phoneme recognition front-end. Secondly, the state decoding of the recognition stage is supported with an uncertainty modeling technique. With the GMM and the uncertainty modeling it is possible to eliminate the need for noise dependent system training. Thirdly, the post-processing of the original method via sinusoidal modeling is replaced with a powerful cepstral smoothing operation. And lastly, due to the improvements of these modifications, it is possible to extend the operational bandwidth of the procedure from 4 kHz to 8 kHz. The performance of the proposed method was evaluated across different noise types and different signal-to-noise ratios. The new method was able to significantly outperform traditional methods, including the one by Xiao and Nickel, in terms of PESQ scores and other objective quality measures. Results of subjective CMOS tests over a smaller set of test samples support our claims.

We present a new method for inventory-style speech enhancement that significantly improves over earlier approaches [1]. Inventory-style enhancement attempts to resynthesize a clean speech signal from a noisy signal via corpus-based speech synthesis. The advantage of such an approach is that one is not bound to trade noise suppression against signal distortion in the same way that most traditional methods do. A significant improvement in perceptual quality is typically the result. Disadvantages of this new approach, however, include speaker dependency, increased processing delays, and the necessity of substantial system training. Earlier published methods relied on a-priori knowledge of the expected noise type during the training process [1]. In this paper we present a new method that exploits uncertainty-of-observation techniques to circumvent the need for noise specific training. Experimental results show that the new method is not only able to match, but outperform the earlier approaches in perceptual quality.

We present a new method for the enhancement of speech. The method is designed for scenarios in which targeted speaker enrollment as well as system training within the typical noise environment are feasible. The proposed procedure is fundamentally different from most conventional and state-of-the-art denoising approaches. Instead of filtering a distorted signal we are resynthesizing a new “clean” signal based on its likely characteristics. These characteristics are estimated from the distorted signal. A successful implementation of the proposed method is presented. Experiments were performed in a scenario with roughly one hour of clean speech training data. Our results show that the proposed method compares very favorably to other state-of-the-art systems in both objective and subjective speech quality assessments. Potential applications for the proposed method include jet cockpit communication systems and offline methods for the restoration of audio recordings.

We are presenting a method for the enhancement of speech in speaker dedicated speech communication systems. The proposed procedure is fundamentally different from most state-of-the-art filtering approaches. Instead of filtering a distorted signal we are re-synthesizing a new ldquocleanrdquo signal based on its likely characteristics. These characteristics are estimated from the distorted signal. We present a successful implementation of the proposed method for a communication system for which speaker enrollment and noise enrollment are feasible. Forty minutes of clean speech training data is usually sufficient for successful denoising. The proposed method compares very favorably to other state-of-the-art systems in both objective and subjective speech quality assessments.

The growing application of caching in Internet applications have heretofore relied largely on qualitative observation and empirical data on the update behavior of Internet data in their design. While it is empirically known that the update behavior of such data is distinctly bimodal, much less is known about the details of these behaviors and the processes that drives them. A detailed study of the composition of modern Web sites that includes in-depth analyses of the changes made to the data of which those sites are composed offers a large potential benefit not only to Internet caching protocol developers but also to a broad cross-section of information technology professionals and researchers. The first results of just such a study are reported in this paper. A variety of popular Internet Web sites were selected and monitored for changes in both composition and content over a period of several weeks. Data collected in this study is then analyzed using traditional stochastic methods. The results of this investigation are summarized and suggested research directions conclude this article

We are proposing a new approach to the solution of the cocktail party problem (CPP). The goal of the CPP is to isolate the speech signals of individuals who are concurrently talking while being recorded with a properly positioned microphone array. The new approach provides a powerful yet simple alternative to commonly used methods for the separation of speakers. It is based on the observation that the estimation of the signal transfer matrix between speakers and microphones is significantly simplified if one can assure that during certain periods of the conversation only one speaker is active while all other speakers are silent. Methods to determine such exclusive activity periods are described and a procedure to estimate the signal transfer matrix is presented. A comparison of the proposed method with other popular source separation methods is drawn. The results show an improved performance of the proposed method over earlier approaches

Quadratic time varying-spectral analysis methods that achieve a high resolution jointly in time and frequency generally suffer from interference terms that obscure the true location of the auto components in the resulting time-frequency representation. Unfortunately, there is no general mathematical model available for an exact distinction between cross-terms and auto-terms. Consequently an attempt to suppress interference can only rely on a few qualitative properties which are commonly associated with cross terms. Most of the reduced interference distributions that have been developed so far exploit the fact that cross terms tend to oscillate and can hence be suppressed by a properly chosen two-dimensional low pass filter. Besides the fact that cross-terms oscillate, they are also known to be responsible for all negative density values of a time-frequency distribution. None of the currently existing methods addresses this characteristic. In this paper we introduce an entirely new approach that achieves a significant interference reduction by specifically exploiting the negative density structure of cross-terms.