Jacquelyn A. Martino

Bio: Jacquelyn A. Martino is an academic researcher from IBM. The author has contributed to research in topics: Business rule & Online community. The author has an hindex of 12, co-authored 56 publications receiving 925 citations.

AI Fairness 360: An Extensible Toolkit for Detecting, Understanding, and Mitigating Unwanted Algorithmic Bias

Abstract: Fairness is an increasingly important concern as machine learning models are used to support decision making in high-stakes applications such as mortgage lending, hiring, and prison sentencing. This paper introduces a new open source Python toolkit for algorithmic fairness, AI Fairness 360 (AIF360), released under an Apache v2.0 license {this https URL). The main objectives of this toolkit are to help facilitate the transition of fairness research algorithms to use in an industrial setting and to provide a common framework for fairness researchers to share and evaluate algorithms. The package includes a comprehensive set of fairness metrics for datasets and models, explanations for these metrics, and algorithms to mitigate bias in datasets and models. It also includes an interactive Web experience (this https URL) that provides a gentle introduction to the concepts and capabilities for line-of-business users, as well as extensive documentation, usage guidance, and industry-specific tutorials to enable data scientists and practitioners to incorporate the most appropriate tool for their problem into their work products. The architecture of the package has been engineered to conform to a standard paradigm used in data science, thereby further improving usability for practitioners. Such architectural design and abstractions enable researchers and developers to extend the toolkit with their new algorithms and improvements, and to use it for performance benchmarking. A built-in testing infrastructure maintains code quality.

AI Fairness 360: An extensible toolkit for detecting and mitigating algorithmic bias

Abstract: Fairness is an increasingly important concern as machine learning models are used to support decision making in high-stakes applications such as mortgage lending, hiring, and prison sentencing. This article introduces a new open-source Python toolkit for algorithmic fairness, AI Fairness 360 (AIF360), released under an Apache v2.0 license ( https://github.com/ibm/aif360 ). The main objectives of this toolkit are to help facilitate the transition of fairness research algorithms for use in an industrial setting and to provide a common framework for fairness researchers to share and evaluate algorithms. The package includes a comprehensive set of fairness metrics for datasets and models, explanations for these metrics, and algorithms to mitigate bias in datasets and models. It also includes an interactive Web experience that provides a gentle introduction to the concepts and capabilities for line-of-business users, researchers, and developers to extend the toolkit with their new algorithms and improvements and to use it for performance benchmarking. A built-in testing infrastructure maintains code quality.

Biometric authentication on a mobile device: a study of user effort, error and task disruption

Abstract: We examine three biometric authentication modalities -- voice, face and gesture -- as well as password entry, on a mobile device, to explore the relative demands on user time, effort, error and task disruption. Our laboratory study provided observations of user actions, strategies, and reactions to the authentication methods. Face and voice biometrics conditions were faster than password entry. Speaking a PIN was the fastest for biometric sample entry, but short-term memory recall was better in the face verification condition. None of the authentication conditions were considered very usable. In conditions that combined two biometric entry methods, the time to acquire the biometric samples was shorter than if acquired separately but they were very unpopular and had high memory task error rates. These quantitative results demonstrate cognitive and motor differences between biometric authentication modalities, and inform policy decisions in selecting authentication methods.

Personalized news retrieval system

Abstract: A video retrieval system is presented that allows a user to quickly and easily select and receive stories of interest from a video stream. The video retrieval system classifies stories and delivers samples of selected stories that match each user's current preference. The user's preferences may include particular broadcast networks, persons, story topics, keywords, and the like. Key frames of each selected story are sequentially displayed; when the user views a frame of interest, the user selects the story that is associated with the key frame for more detailed viewing. This invention is particularly well suited for targeted news retrieval. In a preferred embodiment, news stories are stored, and the selection of a news story for detailed viewing based on the associated key frames effects a playback of the selected news story. The principles of this invention also allows a user to effect a directed search of other types of broadcasts as well. For example, the user may initiate an automated scan that presents samples of broadcasts that conform to the user's current preferences, akin to directed channel-surfing.

Location-based action triggering system and method

Abstract: A target receiver may receive a location-related specification and a time period-related specification, the time period-related specification specifying time period in which to check for the location-related specification. A location monitor may determine current location-related information. A time period monitor may determine current time-related information. An analyzer may determine whether a pair of the location-related specification and the time-period related specification is satisfied by the current-location information and the current time-related information. The analyzer may further provide notification as to whether the pair of the location-related specification and the time-period related specification is satisfied by the current-location information and the current time-related information.

A Survey on Bias and Fairness in Machine Learning

Abstract: With the widespread use of AI systems and applications in our everyday lives, it is important to take fairness issues into consideration while designing and engineering these types of systems. Such systems can be used in many sensitive environments to make important and life-changing decisions; thus, it is crucial to ensure that the decisions do not reflect discriminatory behavior toward certain groups or populations. We have recently seen work in machine learning, natural language processing, and deep learning that addresses such challenges in different subdomains. With the commercialization of these systems, researchers are becoming aware of the biases that these applications can contain and have attempted to address them. In this survey we investigated different real-world applications that have shown biases in various ways, and we listed different sources of biases that can affect AI applications. We then created a taxonomy for fairness definitions that machine learning researchers have defined in order to avoid the existing bias in AI systems. In addition to that, we examined different domains and subdomains in AI showing what researchers have observed with regard to unfair outcomes in the state-of-the-art methods and how they have tried to address them. There are still many future directions and solutions that can be taken to mitigate the problem of bias in AI systems. We are hoping that this survey will motivate researchers to tackle these issues in the near future by observing existing work in their respective fields.

Audiovisual information management system

Abstract: A system, may include, at least one of audio, image, and a video comprising a plurality of frames. A usage preferences description, describing preferences of a user with respect to the use of at least one of the audio, image, and video, where the description normally includes multiple preferences. In one aspect, a protection attribute with respect to at least one of the preferences indicates whether one of the preferences is considered public or private. Other aspects of the user preferences description include other attributes, alone or in combination.

Information processing apparatus, information processing method and program

Abstract: A method is provided for modifying an image. The method comprises displaying an image, the image comprising a portion of an object; and determining if an edge of the object is in a location within the portion. The method further comprises detecting movement, in a member direction, of an operating member with respect to the edge. The method still further comprises moving, if the edge is not in the location, the object in an object direction corresponding to the detected movement; and modifying, if the edge is in the location, the image in response to the detected movement, the modified image comprising the edge in the location.

Media content descriptions

Abstract: A media content description system receives media content descriptions from one or more metadata providers. The media content description system associates each media content description with the metadata provider that provided the description. The media content description system may generate composite descriptions based on the received media content descriptions. The media content description system provides data associated with the media content descriptions to one or more program data providers.

User profile and geolocation for efficient transactions

Abstract: Techniques for providing friction-free transactions using geolocation and user identifiers are described herein. These techniques may ascertain a user's location based on a location of a mobile device. A transaction between the user and a merchant may be completed with zero or minimal input from the user based on the geolocation of the mobile device and the user identifiers. In some implementations, a transaction initiated earlier is completed when the mobile device arrives at the merchant. Additionally, a parent-child or similar relationship may be established between multiple devices. Security on the mobile device based may be provided by biometric identification and calculation of variance from regular movement patterns. Advertisements may be sent to the mobile device based on bids from merchants near to the mobile device. Promotions may be sent to the mobile device when more than a threshold number of mobile devices are located at the same merchant.