Deepfake generation techniques have developed at a rapid rate, making it possible to generate highly realistic yet misleading videos with potentially far-reaching implications for privacy, security, and public confidence. This paper presents a study on the detection of deepfakes using the temporal consistency of facial expression transitions. Our method captures and integrates significant spatial and temporal information, facial edges, and dense optical flow with an Xception-based CNN and a bidirectional LSTM (BiLSTM) with an attention mechanism. We evaluated the approach on a multi-expression dataset obtained from DeeperForensics-1.0, comparing performance systematically across a range of expressions from Angry to Neutral. The experiments demonstrate a detection rate of up to $98.38 \%$ on the combined multi-expressions and point to the unique challenge of less expressive emotions. The findings affirm that face expression continuity examination plays an important part in enhancing the robustness of deepfake detection, achieving a scalable and adaptive approach to verifying the integrity of real-world media.

Visual speaker authentication (VSA) remains vulnerable to sophisticated spoofing methods, such as deepfakes. Traditional deep learning approaches require extensive userspecific enrollment data and show poor generalization to new speakers. In response, we employ a few-shot meta-learning technique, specifically Model-Agnostic Meta-Learning (MAML), integrated with dynamic lip movement analysis utilizing optical flow to develop a scalable anti-spoof VSA framework. We validated our model using the GRID audiovisual dataset, with spoofing attacks simulated via Wav2Lip for deepfake lip synchronization. The results demonstrate the model’s superior performance, evidenced by near-perfect classification accuracy and negligible error rates, addressing two key challenges of VSA: eliminating extensive training data while enabling rapid adaptation to unfamiliar speakers. Our approach significantly surpasses standard non-meta-learning approaches, substantiating its ability to address real-world scenarios.

The availability of low-cost brain-computer interfaces and related software has enabled application developers to access this technology seamlessly. However, unsupervised access to users' brain signals raises alarms for EEG data privacy and identity protection. This paper explores a new direction to selectively anonymize a person's brain signals resulting from a response to a stimulus. These time-locked potentials containing sensitive user information are masked to allow for the intended task/event classification (brain task activity) while minimizing the accuracy of subject classification (brain identity activity).We study the feasibility of an autoencoder architecture, enveloped with regularizers and a multi-objective loss function, to achieve an optimal utility-privacy trade-off for EEG data application. We observed a drop of 35% in the accuracy of the subject's classification, while suffering only a loss of 14% in the accuracy of the task classification. This algorithm can be applied to multi-channel and multi-subject scenarios, and our results demonstrate a proof-of-concept that we can generalize an anonymizing autoencoder architecture to be applicable to intricate stochastic data such as EEG.

In this paper, we have discussed the reproducibility workflow of published Brain-Computer Interface research articles and remarked on the same by replicating two papers having multiple similarities, starting from the same dataset to the classification stages. We followed a step-by-step approach while replicating the work and documenting the assumptions and interpretations made during the process. Finally, we compared the results and discussed how the documentation in BCI research has evolved over 20 years. Through trial and error implementations and calculated deductions, this paper helps determine the importance and relevance of proper documentation, efficient workflows, and the pressing need for direction-specific information flow in the growing field of Brain Computing Interface applications.

The IEEE Baltimore Section sponsored B2CI 2019, the collegiate Brain to Computer Interface competition, to strengthen their relationships with local university and college engineering departments. A primary goal was to use the processing of brain waves as a mechanism for exciting students about electrical engineering principles and the systems engineering build approach. An ancillary goal was to raise student awareness as to the limitations of computer games for disabled players. The competition was designed to motivate students by providing three alternative events in which to compete. One of the events was the brain-controlled computer game contest. Universities were invited to organize teams to design and code games that utilized EEG signals from a headset as a significant part of the game play. This paper describes the EEG technology, details on the brain-controlled computer game event, and lessons learned for future competitions.

Conventional authentication methods utilizing alphanumeric username and passwords, PIN numbers, or any combination thereof have many weaknesses. On modern smart phones there are multiple ways users can authenticate. The traditional username password combination is used often as well as PIN passwords and pattern based passwords. The problem with these methods is that they are still vulnerable. A short password or to a much greater extent a PIN, or a pattern password can be defeated by various techniques such as smudge attacks, key loggers and so on. Our aim with Pass-Pic is to implement a picture based authentication system that is both more secure and easier for the user to both input and remember.

The intention of an Android application, determined by the source code analysis is used to identify potential maliciousness in that application (app). Similarly, it is possible to analyze the unintentional behaviors of an app to identify and reduce the window of vulnerabilities. Unintentional behaviors of an app can be any developmental loopholes such as software bugs overlooked by a developer or introduced by an adversary intentionally. FindBugsTMand Android Lint are a couple of tools that can detect such bugs easily. A software bug can cause many security vulnerabilities (known or unknown) and vice-versa, thus, creating a many-to-many mapping. In our approach, we construct a matrix of mapping between the bugs and the potential vulnerabilities. A software bug detection tool is used to identify a list of bugs and create an empirical list of the vulnerabilities in an app. The many-to-many mapping matrix is obtained by two approaches - severity mapping and probability mapping. These mappings can be used as tools to measure the unknown vulnerabilities and their strength. We believe our study is the first of its kind and it can enhance the security of Android apps in their development phase itself. Also, the reverse mapping matrix (vulnerabilities to bugs) could be used to improve the accuracy of malware detection in Android apps.