Digital Commons Network^™

Anomaly Detection On Multidimensional Time Series, Michael Yeh Visa, Xin Dai Visa, Yan Zheng Visa, Zhongfang Zhuang Visa, Laing Wang Visa, Yujie Fan Visa, Junpeng Wang Visa, Prince Osei Aboagye Visa, Uday Singh Saini Visa, Wei Zhang Visa

Defensive Publications Series

The present disclosure describes a method and system for detecting anomalies in multidimensional time series data using multidimensional matrix profiles are disclosed. A computer system can generate a multidimensional matrix profile corresponding to a multidimensional time series using a pre-sorting or post-sorting technique. The computer system can detect anomalies in the multidimensional time series using the multidimensional matrix profile, e.g., using thresholding or machine learning. The computer system can issue an alert to a requestor if an anomaly has been detected. Additionally disclosed is a k-nearest-neighbour extension to the multidimensional matrix profile anomaly detection methods.

Artificial Intelligence - Comprehensive Guide For Effective Algorithm Evaluation And Sox – Itgc And Business Audit Toolkit And Examples, Ramkumar Bharathan

Defensive Publications Series

Abstract

Artificial intelligence (AI) is rapidly transforming various sectors, raising concerns about its ethical implications and potential risks. AI audits have emerged as a crucial tool for organizations to ensure responsible AI development and deployment. This research report provides a comprehensive guide to conducting AI audits, covering key aspects such as how to conduct an audit, how to use AI audit toolkits, types of AI audits, legal and regulatory requirements, ethical considerations, and available resources. By utilizing this toolkit, organizations can effectively assess and mitigate risks associated with AI, promoting fairness, transparency, and accountability in their AI systems.

Appendix - …

Artificial Intelligence Powered Brand Protection Tool

Defensive Publications Series

Methods, systems, or apparatuses for analyzing threats to a brand that enables the use of artificial intelligence to analyze one or more electronic publications for unauthorized product listing, or generate a report of product listings associated with information provided in a query. The disclosed system may receive one or more queries, search one or more electronic publications, obtain data from the one or more electronic publications, create a database of the obtained data from the one or more electronic publications, analyze the obtained data, or generate one or more reports with information related to the obtained data.

Enhanced Calendar Interface Using Ai For Visualizing Meeting Relationships, Ajay Prasad

Defensive Publications Series

Current calendar applications lack the ability to effectively visualize relationships between interconnected meetings and do not provide an easy-to-use interface to view the progression of project-related meetings over time or to see the relationships between various meetings. This disclosure describes an enhanced calendar application and associated user interface powered by artificial intelligence (AI) that can automatically identify and visually represent relationships between meetings. With user permission, AI techniques are used to analyze meeting data and/or metadata to determine if two meetings are related. Related meetings are displayed via visual elements such as connecting lines, colors, tooltips, etc. to enable users …

Augmenting Calendar View With Visual Representations Of Event Information, Gulmohar Khan, Ramprasad Sedouram

Defensive Publications Series

Visual representations in current calendar applications provide minimal to no event information beyond the time slot, title, and location. For effective scheduling and time management, users often need to know additional event information which can be accessed only by opening and consulting each calendar event separately. This disclosure describes techniques to address these shortcomings by augmenting calendar views with richer visual depictions of individual events to enable users to view contextually relevant event-related information at a glance. With appropriate user permissions, the richer depictions are provided by overlaying multiple relevant pieces of data and interactive elements in the visual depiction …

Artificial Intelligence To Quantify Employee Communication And Personality Dynamics

Defensive Publications Series

Systems and methods for evaluating an employee’s professional personality are provided. The systems and methods may enable the use of artificial intelligence to perform sentiment analysis on one or more employee communications, and quantify one or more personality characteristics based on the one or more employee communications. The systems and methods may compare the quantified personality characteristics to one or more employees, and may generate one or more employee scores for employees for each of the one or more personality characteristics.

Method To Generate Electric Power Using Seebeck Effect On A Flexible Pipe, Baker Hughes Company

Defensive Publications Series

The thermoelectric effect is the direct conversion of temperature differences to electric voltage and vice versa via a thermocouple. A thermoelectric device creates a voltage when there is a different temperature on each side. This effect can be used to generate electricity.

Auditing The Future: Ensuring Ethical And Secure Ai And Quantum Computing Itgc And Business Controls, Ramkumar Bharathan

Defensive Publications Series

Artificial intelligence (AI) and quantum computing are rapidly transforming the business landscape, demanding a parallel evolution in auditing frameworks to ensure their ethical, secure, and reliable deployment. This paper presents a comprehensive auditing approach tailored to the unique challenges posed by these technologies, aligning with Sarbanes-Oxley (SOX), International Financial Reporting Standards (IFRS), and Generally Accepted Accounting Principles (GAAP). It addresses key control areas, including access controls, change management, data integrity, model validation, and incident management, providing detailed audit steps with walkthroughs and testing examples. The framework incorporates EU guidance on algorithms and AI governance, emphasizing algorithmic bias mitigation, data privacy, …

Providing Enhanced Learning In Llms By Associating Data With Role-Based Source Information, Laxmi Mukund, Shraddha S K

Defensive Publications Series

In today's fast-paced world, the rapid introduction of new product features necessitates accurate and timely documentation. Traditional manual documentation methods are inefficient, error-prone, and often result in outdated information. Techniques are presented herein to address these challenges by leveraging Large Language Models (LLMs) to automate the generation of precise and reliable documents by associating data with well-defined role-based source information. The impact of this solution goes beyond generating documentation for new features; it can also be extended to enable targeted stakeholder interactions, to identify speakers in recorded conversations, and to classify unclassified dialogues.

System And Method To Authorize A Payment Transaction Using A Digital Glove, Sri Sai Krishna Kiran Atmuri

Defensive Publications Series

The present disclosure relates to a system and method to authorize a payment transaction using a digital glove. The present disclosure suggests tapping a cardholder’s digital glove on a merchant payment system without entering a Personal Identification Number (PIN) to initiate a payment transaction. The merchant payment system transmits the payment transaction data to an issuer system through a network via an acquirer server for authorization. Upon notification, the cardholder enters the PIN by performing a sequence of pinches between a thumb finger distal phalange and the other finger phalanges of the digital glove. Thereafter, encrypt the chronological order of …

Thematic Navigation Using Generative Artificial Intelligence And Augmented Reality, Ajay Prasad

Defensive Publications Series

Navigation apps offer accurate routes, but they are purely utilitarian, offering little scope for user engagement and enjoyment. This disclosure describes techniques of navigational experience that leverage generative AI (gen-AI) and augmented reality to transform routine navigation tasks into thematic, immersive, and memorable journeys. Users can select from a variety of themes from popular culture, obtained with permission from the content creators, to overlay the navigation interface and the real-world environment with relevant content. Gen-AI techniques are used to create themed content, renaming landmarks, generating background visuals, creating audio cues and directions, etc. Thematic elements can be projected onto the …

Region-Based Cross-Component Prediction Parameter Derivation, Xiang Li, Debargha Mukherjee

Defensive Publications Series

Region-based cross-component prediction parameter derivation improves convolutional cross-component mode (CCCM) prediction by enabling filter coefficients for predicting chroma samples from luma samples to be derived for an entire region of a frame of a video stream, such as a coding tree unit (CTU), rather than requiring that such filter coefficients be derived for each individual coding unit (CU). Deriving the filter coefficients for an entire region instead of for each individual CU under processing significantly reduces the latency in video coding and thus enables CCCM prediction to be used in hardware coders. The filter coefficients for the region are derived …

Techniques To Facilitate Hitless Recovery From A Flash Single Event Upset (Seu) Error Through Virtualization, Jay Yoo, Mike Thornburg, Mark Chu, Daniel I Walker, Vinod Cherukatt, Son P Le, Jason F Haase, Ramesh Yeevani, Manish Agarwal

Defensive Publications Series

Many embedded systems use an embedded MultiMediaCard (eMMC) as the main storage device for such systems. When the eMMC for an embedded system is connected to the host Central Processing Unit (CPU) through an intermediate device, such as a Universal Serial Bus (USB) bridge, recovery from a transient eMMC failure, such as a single-event upset (SEU), is not available and such a failure could bring down an entire network. For example, conventional recovery processes for a system that experiences such a transient failure due to an SEU issue typically involve performing a system power cycle, which can cause significant network …

Benefits Marketplace, Ana Luisa Wu, Jenny Ramirez, Jose Ricardo Tafur Jr, Ricardo Collada, Maikel Zuniga, Manuel Ferreiro Caballero, Salvador Ricardo, Liset Iturriaga Chavez, Francisco Alberdi, Alexander Dale, Orlando Quesada, Ernesto Garcia Vazquez, Daniel Astudillo, Laura Figueroa, Sebastian Hernandez, Cristina Villarroel, Yuset Rosello Moya, Rachel Quijano Vizoso

Defensive Publications Series

A method of provisioning a benefits marketplace platform is disclosed. According to the method, a payment network server computer presents payment network benefits to a consumer and receives feedback from the consumer based on the payment network benefits presented to the consumer. A profiling module profiles the payment network benefits based on the feedback received from the consumer and provides a benefits profile to the payment network server computer. The payment network server provides a recommendation of benefits to the consumer based on the benefits profile. A benefits module receives benefits selected by the consumer from a checkout function, applies …

Ai-Based Detection And Formalization Of In-Session Annotations In Virtual Meetings, Xun Qian, Ruofei Du

Defensive Publications Series

This disclosure describes techniques to automatically detect annotations such as freeform sketches and text captions provided by users during online video conferences and generate formalized representations of the annotations for in-meeting display as well as for subsequent retrieval. The techniques can be implemented for video conferences, including where participants participate via extended reality (XR) devices. Implementation of the techniques enables users to create high-fidelity sketches and annotations while participating in a video conference (virtual meeting). With user permission, an artificial intelligence (AI) module is utilized that detects user speech, a targeted first-person view, and user sketches provided during a virtual …

Automatic Generation Of Search Dimensions For Query Refinement, Manasa Babu, Sanika Barve, Ramprasad Sedouram

Defensive Publications Series

This disclosure describes techniques to seamlessly refine and update search queries, resulting in an enhanced user experience and convenience. Per the techniques, in addition to traditional textual refining of queries, users can fine-tune the output results for specific requirements pertaining to information size, personalization, multi-device display of search results, etc. A user can adjust the granularity of the search results, from a gist to in-depth or personalized knowledge, by simple gestures such as tilting the device towards or away from themselves. Information size in the search results can be automatically adjusted to suit the present display size. Search results can …

Predictive Flow Capture Controls For Multi-Domain Network Assurance Tools, Robert Barton, Jerome Henry, Abhishek Kumar, Gino Corleto

Defensive Publications Series

Advanced network assurance tools provide for the ability to manage network assurance across a variety of service domains. Such tools may facilitate synthetic network monitoring through active monitoring of live flow data. While this opens new opportunities for network assurance, architecturally it also presents challenges related to the orchestration of how and where flow reporting is established and managed. Techniques are proposed herein that include providing an AI/ML mechanism to predict network problems such that flow collection can be automated to be performed only where necessary, when necessary, and for the protocols and/or hosts that are experiencing issues. The techniques …

Prevention Of Internet Protocol Address Reusability-Based Domain Takeover, Pierre Precourt, Patrick Spiegel

Defensive Publications Series

For each domain of an organization a corresponding IP address is identified. For each corresponding IP address, its registration, ownership information, and/or the corresponding media access control (MAC). Optionally, whether the IP address is allocated to an existing host is determined. For each IP address, address allocation requests are repeatedly sent until the IP address of interest is allocated in response to one of the requests. Alternatively, an API reporting the address is available is queried to determine whether the IP address of interest is currently allocated or available for allocation. As a result, cloud service providers can mitigate the …

Artificial Intelligence Based Power Saving Solutions For 5g Satellite Coverage In Remote Areas, Snezana Mitrovic, Vinay Saini, Alwin Xavier, Wojciech Grobel

Defensive Publications Series

Satellites can facilitate network coverage for Internet of Things (IoT) -based sensors in remote regions, eliminating the need for ground-based infrastructure. However, this advancement introduces new obstacles, particularly concerning the financial costs of satellite-based network coverage and the power efficiency of IoT devices. Proposed herein are techniques to implement an artificial intelligence (AI)/machine learning (ML) module within the 5G network infrastructure that intelligently adapts power-saving settings for IoT devices to align with the timing of satellite availability, optimizing device energy consumption while maintaining reliable data transmission.

Thermal Imaging Based Solutions For Implementing Platform Management In Networking/Server Equipment, Hari Pillai, Phong Ho, Stephen A Scearce, Lauren A Marino

Defensive Publications Series

Existing systems for internal monitoring of networking/server equipment are not scaling to meet the needs of the next generation of electronic system issues. Current solutions for internal monitoring only provide a limited amount of information that leaves gaps in the understanding of the condition and integrity of the networking products. To overcome these issues, proposed herein are techniques that leverage real time infrared (IR) monitoring to quickly detect intrusive events, thermal complications, corrosion, and dust accumulation before damage to the networking products or customers occur.

Establishing A Secure Sharing Environment Using Partial Cryptographic Keys, Salmaan Rashid

Defensive Publications Series

A system for establishing a secure sharing environment using partial cryptographic keys. The proposed system uses a complete cryptographic key divided into multiple partial keys and provided to two or more participants. The participants attest each TEE of the secure sharing environment desired for use and send, to each TEE, their respective partial key. Each TEE combines the received partial keys to generate the complete cryptographic key. Using the complete cryptographic key, the TEEs can securely communicate and share data with each other.

Automated Regression Testing Framework For Large Language Models, Claudio Criscione, Sebastian Lekies

Defensive Publications Series

This disclosure describes a system for automated regression testing of large language models (LLMs). LLMs can be trained or fine-tuned with safety-oriented data and similar techniques to prevent undesirable outputs. However, additional training and fine-tuning can result in an LLM regressing and producing undesirable outputs. These regressions can be difficult to detect and track given the fast pace of research and development in this field. This system provides automated infrastructure and techniques for testing new versions of LLMs for regression against stored tests. The system includes integrator modules that provide a uniform testing interface for various types of LLMs and …

Defined-Trust Transport Protocol For Limited Domains, Kathleen Nichols

Defensive Publications Series

The Defined-trust Transport (DeftT) is designed to provide secure, default-deny networking for closed communities with dynamic membership and a collection-based construct that is efficient on broadcast media. Such closed communities, also known as limited domains [LDCCR], are frequently used in Operational Technology (OT) networking, in particular , in Critical Infrastructure uses. DeftT is designed to express and enforce application- and deployment-specific integrity, authentication, access control and behavior constraints directly in its protocol modules. It enables secure and completely self-contained (e.g., no external identity servers or certificate authorities) overlay networks where credentialed members can join and leave at any time. Security …

Action Item Management Across Virtual Collaboration Meetings And Spaces, Chidambaram Arunachalam, Gonzalo A Salgueiro, Paul Anholt, Hazim Dahir

Defensive Publications Series

The concept of sharing key points and identifying action items based on a discussion in a virtual collaboration meeting is the most effective way of documenting the decisions made and tasks assigned during the meeting. However, meeting summaries including AI based summaries, does not provide information on whether the action items were already completed during the meeting. Proposed herein are techniques to determine completion of action items related to a specific topic based on conversations, real-time media, file share events across related virtual collaboration spaces and virtual meetings.

Closed-Loop Temperature Controller With Temperature Prediction

Defensive Publications Series

The present disclosure generally relates to closed-loop controllers, and more particularly, to a closed-loop temperature controller with temperature prediction

Capacitor Touch Interface For Ai Interaction, Hp Inc

Defensive Publications Series

This disclosure presents the Capacitor Touch Interface for AI Interaction, a system that extends touch-sensitive technology beyond traditional display surfaces to enable more intuitive device interaction. By integrating capacitor sensors across various laptop surfaces and coupling them with AI capabilities, the system allows users to activate their device and trigger AI assistance through a simple touch gesture. Upon touch detection, the system automatically activates a microphone to capture spoken prompts, which are then transcribed and processed to execute AI-driven actions. Our approach focuses on hardware-firmware integration to ensure reliable touch detection and audio capture while maintaining low latency. Initial …

Quantum And Bio-Inspired Self-Healing Networks, Na

Defensive Publications Series

This disclosure describes techniques to reduce the costs and inefficiencies associated with providing redundancy in networks by using self-healing network architectures. The techniques feature quantum-inspired routing algorithms (QIRA) and bio-inspired self-healing (BISHA). QIRA leverages quantum principles to concurrently explore multiple paths and to share information about network conditions through quantum entanglement. BISHA leverages cellular automata to mediate inter-device communication, such that when a failure is detected, the device is isolated and information about the failure is communicated to neighboring devices. The neighboring devices collaborate to reconfigure the network topology and reroute data traffic around the failed component. Machine learning algorithms, …

Data-Driven Design For Thermal Management And Control In Datacenters, Na

Defensive Publications Series

Design of thermal controllers for datacenter machines has traditionally relied on heuristics and trial-and-error testing, resulting in lengthy design times and possibly suboptimal controller parameters. This disclosure describes techniques that leverage control and linear-system theory to build models that enable the software simulation of thermal controllers used in datacenters. Per the techniques, test data is captured from a real machine and used to create a reduced-order model of the machine and its thermal controller. A transfer function of the reduced-order model is obtained using linear system theory. Control theory is leveraged to determine the appropriate controller parameters for the machine …

Real-Time Message Sentiment Augmentation By Emoji Symbols, Gregory Savage, Alex Yung

Defensive Publications Series

Aspects of the disclosure are directed to real-time message sentiment augmentation by emojis. A large language model (LLM) can determine, for each message that is shared in a virtual chat, a presumed sentiment of an author of a message and a presumed audience reaction toward the latest message in the virtual chat. The LLM can identify an emoji that captures the presumed sentiment of the author. The LLM-identified emoji can be visually associated with an icon/avatar that corresponds to the author. The LLM can identify emojis that capture the presumed audience reaction toward the latest message in the virtual chat. …

A Method And A System For Clustering Similar Transaction Disputes, Rupali Shah

Defensive Publications Series

The present disclosure provides a method and a dispute clustering system for Clustering similar transaction disputes. In the present disclosure, during a training phase, a dispute clustering system may receive historical transaction disputes to form sets of historical transaction disputes with similar causes. Each set of historical transaction disputes may be resolved using the same resolving method. In real time, the dispute clustering system may receive one or more transaction disputes. Then, the dispute clustering system may determine a related set of historical transaction disputes for each of the one or more transaction disputes by comparing the causes and rules …