Training deep graph neural networks (GNNs) poses a challenging task, as the performance of GNNs may suffer from the number of hidden message-passing layers. The literature has focused on the proposals of over-smoothing and under-reaching to explain the performance deterioration of deep GNNs. In this paper, we propose a new explanation for such deteriorated performance phenomenon, mis-simplification, that is, mistakenly simplifying graphs by preventing self-loops and forcing edges to be unweighted. We show that such simplifying can reduce the potential of message-passing layers to capture the structural information of graphs. In view of this, we propose a new framework, edge enhanced graph neural network (EEGNN). EEGNN uses the structural information extracted from the proposed Dirichlet mixture Poisson graph model (DMPGM), a Bayesian nonparametric model for graphs, to improve the performance of various deep message-passing GNNs. We propose a Markov chain Monte Carlo inference framework for DMPGM. Experiments over different datasets show that our method achieves considerable performance increase compared to baselines.

The abstracts of scientific papers typically contain both premises (e.g., background and observations) and conclusions.Although conclusion sentences are highlighted in structured abstracts, in non-structured abstracts the concluding information is not explicitly marked, which makes the automatic segmentation of conclusions from scientific abstracts a challenging task.In this work, we explore Normalized Mutual Information (NMI) as a means for abstract segmentation.We consider each abstract as a recurrent cycle of sentences and place two segmentation boundaries by greedily optimizing the NMI score between the two segments, assuming that conclusions are strongly semantically linked with preceding premises.On nonstructured abstracts, our proposed unsupervised approach GreedyCAS achieves the best performance across all evaluation metrics; on structured abstracts, GreedyCAS outperforms all baseline methods measured by P k .The strong correlation of NMI to our evaluation metrics reveals the effectiveness of NMI for abstract segmentation.1

Research efforts tried to expose students to security topics early in the undergraduate CS curriculum. However, such efforts are rarely adopted in practice and remain less effective when it comes to writing secure code. In our prior work [18], we identified key issues with the how students code and grouped them into six themes: (a) Knowledge of C, (b) Understanding compiler and OS messages, (c) Utilization of resources, (d) Knowledge of memory, (e) Awareness of unsafe functions, and (f) Understanding of security topics. In this work, we aim to understand students' knowledge about each theme and how that knowledge affects their secure coding practices. Thus, we propose a modified SOLO taxonomy for the latter five themes. We apply the taxonomy to the coding interview data of 21 students from two US R1 universities. Our results suggest that most students have limited knowledge of each theme. We also show that scoring low in these themes correlates with why students fail to write secure code and identify possible vulnerabilities.

Current variational inference methods for hierarchical Bayesian nonparametric models can neither characterize the correlation structure among latent variables due to the mean-field setting, nor infer the true posterior dimension because of the universal truncation. To overcome these limitations, we propose the conditional and adaptively truncated variational inference method (CATVI) by maximizing the nonparametric evidence lower bound and integrating Monte Carlo into the variational inference framework. CATVI enjoys several advantages over traditional methods, including a smaller divergence between variational and true posteriors, reduced risk of underfitting or overfitting, and improved prediction accuracy. Empirical studies on three large datasets reveal that CATVI applied in Bayesian nonparametric topic models substantially outperforms competing models, providing lower perplexity and clearer topic-words clustering.

Often, security topics are only taught in advanced computer science (CS) courses. However, most US R1 universities do not require students to take these courses to complete an undergraduate CS degree. As a result, students can graduate without learning about computer security and secure programming practices. To gauge students' knowledge and skills of secure programming, we conducted a coding interview with 21 students from two R1 universities in the United States. All the students in our study had at least taken Computer Systems or an equivalent course. We then analyzed the students' approach to safe programming practices, such as avoiding unsafe functions like gets and strcpy, and basic security knowledge, such as writing code that assumes user inputs can be malicious. Our results suggest that students lack the key fundamental skills to write secure programs. For example, students rarely pay attention to details, such as compiler warnings, and often do not read programming language documentation with care. Moreover, some students' understanding of memory layout is cursory, which is crucial for writing secure programs. We also found that some students are struggling with even the basics of C programming, even though it is the main language taught in Computer Systems courses.

High linearity/sensitivity and a wide dynamic sensing range are the most desirable features for pressure sensors to accurately detect and respond to external pressure stimuli. Even though a number of recent studies have demonstrated a low-cost pressure sensing device for a smart insole system by using scalable and deformable conductive materials, they still lack stretchability and desirable properties such as high sensitivity, hysteresis, linearity, and fast response time to obtain accurate and reliable data. To resolve this issue, a flexible and stretchable piezoresistive pressure sensor with high linear response over a wide pressure range is developed and integrated in a wearable insole system. The sensor uses multi-walled carbon nanotubes and polydimethylsiloxane (MWCNT/PDMS) composites with gradient density double-stacked configuration as well as randomly distributed surface microstructure (RDSM). The randomly distributed surface of the MWCNT/PDMS composite is easily and non-artificially generated by the evaporation of residual IPA solvent during a composite curing process. Due to two functional features consisting of the double-stacked composite configuration with different gradient MWCNT density and RDSM, the pressure sensor shows high linear sensitivity (∼82.5 kPa) and a pressure range of 0-1 MPa, providing extensive potential applications in monitoring human motions. Moreover, for a practical wearable application detecting the user's real-time motions, a custom-designed output signal acquisition system has been developed and integrated with the insole pressure sensor. As a result, the insole sensor can successfully detect walking, running, and jumping movements and can be used in daily life to monitor gait patterns by virtue of its long-term stability.

Introductory computer science courses, such as Computer Systems, could be used to provide the first exposure to computer security to students. However, prior work has shown that, in the US's top R1 universities, computer systems courses are not taught with security in mind. It was also shown that students and instructors use unsafe functions in their code, leading to security vulnerabilities. In this paper, we focused on the textbooks used for computer systems courses. We analyzed the discussion of security topics and the use of unsafe functions in the thirteen textbooks used in the top 30 R1 universities in the US for teaching computer systems. We show that many textbooks do not discuss security at all, while some limit their discussion to "undefined behavior'', ignoring that opportunity to discuss potential security issues associated with the undefined behavior. Furthermore, textbooks that talk about security continue using unsafe functions throughout (though not necessarily in vulnerable ways but also without any warning or explanation). We also show that many textbooks do not warn about unsafe functions they use or teach how to use them safely.

Introductory computer systems courses teach students how a single program is executed inside a computer, providing them with their first exposure to the logical internals of computing systems. This is one of the first introductory courses where students can learn about security and the need for robust coding. However, currently, these courses are taught with a focus on functionality and efficiency only, ignoring security almost entirely.