As computer science is integrated into a wider variety of fields, block-based programming languages like Snap!, which assemble code with visual blocks rather than text syntax, are increasingly used to teach computational thinking (CT) to students from diverse backgrounds. Although automated evaluators (autograders) for programming assignments usually focus on runtime efficiency and output accuracy, effective evaluation of a student's CT skills requires assessing coding best practices, such as decomposition, abstraction, and algorithm design. While autograders are commonplace for text languages like Python, we present a machine learning approach to assess how effectively block-based code demonstrates understanding of CT fundamentals. Our dataset consists of Snap! programs written by students new to coding and evaluated by instructors using a CT rubric. We explore how to best transform these programs into low-dimensional features to allow encapsulation and repetition patterns to emerge. Experimentation involves comparing the effectiveness of a suite of clustering models and similarity metrics by analyzing how directly automated feedback correlates to the course staff's manual evaluation. Lastly, we demonstrate the practical application of the autograder in a classroom setting and discuss scalability and feasibility in other domains of CS education.

To accelerate the adoption of computational thinking (CT), we have developed TuneScope, an online platform for introducing novices to programming in the context of music. TuneScope combines a sound analysis & synthesis tool with Snap!, a computing language developed at the University of California, Berkeley. This demo explores CT concepts such as decomposition, patterns, abstraction, and algorithms in TuneScope while also exploring the creation of four cascading musical components from (1) sequences of notes, (2) musical chords, (3) sampled sounds, and (4) synthesized sounds. The challenge is to design activities that include authentic music learning as well as genuine computational thinking. In this demo, we show concepts around sequence (the order in which musical notes appear in time; and the order of statements in a computer program) and repetition (includes repeats as well as the structure of melodies; and computing loops and recursion). The instructional activities in this demo have been piloted three times in an associated course at the University of Virginia. Data collected from the course suggest a positive effect on both the understanding of CT concepts and the comprehension of music. More detail on TuneScope can be found at https://maketolearn.org/tunescope/.

Machine Learning (ML) has become one of the highly participated courses at the undergraduate level in Computer Science. Open-source ML libraries make it easy for students to implement papers, share ideas, and conduct experiments on large scale datasets. With the emergence of public dataset portals (such as Kaggle, Amazon Datasets, and Google Datasets Search), the open-source community has produced many useful, high-quality libraries (such as Scikit-Learn, PyTorch, Keras, and Tensorflow among others). These open-source tools aim to make state-of-the-art ML algorithms and large-scale datasets accessible to all. While these ML libraries and datasets can benefit many undergraduate students in their pursuit of data-related careers, the task of choosing them for instructional purposes can be daunting for two reasons. First, all of these tools have advantages, drawbacks, and many overlapping issues. There is no single tool or dataset that covers all of the ML instructional needs. Second, due to the rapid advancements in the field, instructors often find a lack of comprehensive guidelines or standards on evaluating the instructional usability and real-world performance of open-source tools. How can these libraries and tools be integrated to aid the instructional activities of both classical machine learning as well as deep learning? This BOF will provide a platform for the discussion of the development of an open-source toolkit to support the teaching and learning of ML at the undergraduate level.

Increasing the representation of minorities in computer science (CS) has become a national priority. One of the many reasons minority students nationwide choose not to study CS is that they often lack mentors and role models to encourage them early in their learning. In her seminal article, Jeannette Wing argues that computational thinking (CT) is an emerging essential skill that should become an integral part of K-12 education. However, a big issue facing this initiative is that many K-12 teachers find themselves lack (1) relevant materials, (2) systematic training, and (3) a supportive community. The democratization of media, such as photos and videos, has provided a great variety of options to educate a broad audience on myriad topics. In this poster, we will describe the challenges and successes of using animated videos including its beauty, soundness, and utility as critical elements in establishing a strong CT comprehension while engaging K-12 teachers in a non-threatening way. Using a process of iterative design, we have found ways to integrate CT concepts in six non-CS disciplines in the K-12 curriculum. The teachers who have collaborated with us respond positively that this design approach provides them with a greater comprehension of the CT concepts while giving them exciting instructional activities. Therefore, this poster may be of interest to any CS educator who wishes to improve the engagement of K-12 teachers while sustaining a CT training program at their institution.

Increasing women's representation in computer science (CS) has become a national priority. One of the many reasons female students nationwide choose not to finish their study in CS is that they do not feel a deep sense of belonging in the major. To foster the sense of belonging among female CS students, the affective learning outcomes, which are adapted from Bloom's Taxonomy on human learning, are integrated into the context of peer tutoring as five successive stages. Through the five stages of affective peer tutoring, students gradually deepen their sense of belonging in CS by: (1) being aware of the tutoring services; (2) proactively seeking answers to their programming questions; (3) recognizing the value of self-efficacy; (4) discussing learning issues in a supportive environment; and (5) contributing to a peer-led learning program to help others. Our data indicate that affective peer tutoring has resulted in an increased level of peer-to-peer interaction outside of the classrooms, significantly better grade performance in introductory programming courses, and improved retention rate among female CS students. Therefore, this poster may be of interest to any CS educator who wishes to improve the interaction, performance, and retention among female CS students while sustaining a peer-led learning program at their institution.

The studies of the network formed by social insects require the motion analysis of their interactions and movements in videos over an extended period of time. Automated detection is an important field of interest because it enables the motion analysis in large-scale experiments. When an automated detection method is applied to various insect types, the training task often involves the collection of a large number of labels provided by human experts. To save the experts' time and effort, unlabeled data have been recently employed to supplement the training. In this paper, we utilize the spatiotemporal connectivity of the unlabeled data to regulate the training of a detector on a new insect type. Our key contribution is integrating the spatiotemporal connectivity among the unlabeled samples to determine the weighting scheme of the existing classifiers from multiple sources. The evaluation on 3 data sets of social insects consisting of 6,000 samples demonstrates that a detector trained using our method can achieve comparable performance to previous approaches while reducing the training labels up to 16 times. Since the proposed method is based on regularizing the unlabeled samples based on their spatiotemporal connectivity, we refer to it as the Spatio-Temporally Regularized Adaptive Learning (STRAL).

Robust inspection is important to ensure the safety of nuclear power plant components. An automated approach would require detecting often low contrast cracks that could be surrounded by or even within textures with similar appearances such as welding, scratches and grind marks. We propose a crack detection method for nuclear power plant inspection videos by fine tuning a deep neural network for detecting local patches containing cracks which are then grouped in spatial-temporal space for group-level classification. We evaluate the proposed method on a data set consisting of 17 videos consisting of nearly 150,000 frames of inspection video and provide comparison to prior methods.

The pollen grains of different plant taxa exhibit various shapes and sizes. This structural diversity has made the identification and classification of pollen grains an important tool in many fields. Despite the myriad of applications, the classification of pollen grains is still a tedious and time-consuming process that must be performed by highly skilled specialists. In this paper, we propose an automatic classification method to discriminate pollen grains coming from a variety of taxonomic types. First, we develop a new feature that captures the spikes of pollen to improve the classification accuracy. Second, we take advantage of the classification rules extracted from the existing pollen types and apply them to the new types. Third, we introduce a new selection criterion to obtain the most valuable training samples from the unlabeled data and therefore reduce the number of needed training samples. Our experiment demonstrates that the proposed method reduces the training effort of a human expert up to 80% compared to other classification methods while achieving 92% accuracy in pollen classification.