As machine learning systems permeate everyday life, the urgent need for safety, fairness, and quality assurance intensifies. Various frameworks exist to guide this process, yet the challenge lies in seamlessly applying these tools in real development scenarios. In our previous work, we introduced a risk assessment tool grounded in the Failure Mode Effect Analysis (FMEA) methodology and some of these existing frameworks. While the tool showed promise in practical industrial settings, it lacked a comprehensive evaluation and actionable strategies to address the risks found. In this paper, we unveil an improved version of this tool, drawing on experiences and insights from diverse case studies in collaboration with an industry partner. The new methodology offers a robust, end-to-end guideline for evaluating machine learning systems based on a comprehensive and practical functionality analysis, risk assessment, and countermeasure planning. Following multiple workshop evaluations, the tool demonstrated its potential to enable safer and more reliable deployment of AI technology, by helping to visualize the complexly correlated issues affecting such systems, and to navigate the tradeoffs when solving them.

The performance of human workers can be fluctuated due to changes in the cognitive state during sustained work. Though past researches have made human performance monitoring possible by utilizing physiological signals, little attention has been paid to the context of office works. This research proposes a multimodal approach to estimate office task performance. A transcription typing experiment was conducted to simulate the real working environment while typing speed and error rate represented as performance metrics. Physiological data collected during the experiment, together with conventional machine learning algorithms showed feasibility to accurately predict two levels (good/bad) of task performance. More importantly, a comprehensive comparison between choices of modality suggests that using data from particular sources could gain predictive performance comparable to the complete set of modalities.

Anonymization is a common technique for publishing a location data set in a privacy-preserving way. However, such an anonymized data set lacks trajectory information of users, which could be beneficial to many location-based analytic services. In this paper, we present a dynamic pseudonym scheme for constructing alternate possible paths of mobile users to protect their location privacy. We introduce a formal definition of location privacy for pseudonym-based location data sets and develop a polynomial-time verification algorithm for determining whether each user in a given location data set has sufficient number of possible paths to disguise the user's true movements. We also provide the correctness proof of the algorithm.

Project ICHIGAN is a voluntary-based attempt to build a reference IT architecture for local governments that can withstand large-scale natural disasters. This architecture is unique in that 1) it has the concept of {em phases} of the situation for which different priorities on non-functional requirements are applied, and 2) the functionalities and services provided by the IT systems in the suffered area will be taken over by those of the ``coupled'' local government. These features pose specific challenges on the information security policy, especially because different policies need to be applied for the different modes. This paper describes two key elements to enable the policy, policy templates and deferred authentication.

The information technology industry in Japan has required universities to provide project management education. In Tokyo University of Technology, role-play training has been carried out as part of project management education. The role-play scenarios necessary to run role-play exercises have been created in accordance with the ADDIE (Analysis, Design, Development, Implementation, Evaluation) model. This paper describes a method for evaluating project management competency that learners gain through role-play training conducted using the scenarios. Competency in project management is assessed from a learner's behavior characteristics in taking an appropriate action when needed. We first examined the quality of the role-play scenarios by using a design checklist based on Goal-Based Scenarios (GBS). In addition, we analyzed the behavior of each learner during a role-play exercise by using rubrics based on how the user behaved. A high correlation was found between the acquired skill with which learners generally played the role assigned to them in role-play training and the level of quality of the role-play scenario. Based on the analysis results, we will propose a method for helping learners to be able to take effective action by providing appropriate advice from a software agent and feedback from a teacher, along with use of the GBS checklist.

The information technology industry has required universities to provide project management education. In Tokyo University of Technology, role-play training has been carried out as a part of project management education. We have been developing an on-line group work training system named PROMASTER (Project Management Skills Training Environment) which can gather the behavior records of each learner to realize PBE (Profile Based Education). Also we have developed a software agent system, named BONAMI (agent system Based ON Aggregated Mentoring and expert Intelligence for project management), which plays the role of an adviser, and monitors the behavior of each learner, when necessary encouraging a learner to take the action. However BONAMI cannot suggest that a learner takes the proper procedures in a timely manner. In this paper, we propose an instruction method for providing advice or a hint in a timely manner. The proposed method is based on the principle that “the correct solution should be derived through the correct orderly manner”. In particular, the method is to extract the conversation patterns from the chat messages and then to compare them to model patterns. According to the result of analyzing the actual conversation patterns, the learners' acquired skill level was found to be positively correlated with the degree of the desirable patterns. This result suggests that the degree of desirable patterns might correspond to the correct orderly approach. Consequently the proposed method encourages learners to discuss in an orderly correct manner to arrive at a correct solution.

This paper describes an on-line group work training environment which integrates a system implementing role-play training for project management education, the development of role-play scenarios, and the monitoring and analysis of the behavior track records of each learner while a role-play is in progress. The integrated environment can motivate a student to study project management and provide an effective means of education. In order to improve the students' skill level through role-play training, time consuming work is required for a faculty staff to carefully develop and review each role-play scenario. The proposed environment could shorten the debug time by an average of 63%, and can monitor and analyze students' learning attitudes as the role-play takes place.

A role-play training system for enhancing the efficiency of education in project management is proposed and its effectiveness evaluated. The system is essentially a Web application that runs an interactive virtual project, in which each learner assumes the role of one of its stakeholders. It is a non-intelligent system driven by a scenario that defines the scope of the virtual project, gives information about the stakeholders, and issues a series of directions to the participants. An institution can improve the quality of education by providing individualized guidance to learners based on an analysis of their learning behavior recorded in their logs.

This paper proposes WS-attestation, attestation architecture on Web services framework. We aim at providing software oriented, dynamic and fine-grained attestation mechanism that leverages TCG technologies to increase trust and confidence in integrity reporting. In addition, the architecture allows efficient binding of attestation with application context, privacy protection, as well as infrastructural support for attestation validation.

Web services are emerging technologies to reuse software as services over the Internet by wrapping underlying computing models with XML. Web services are rapidly evolving and are expected to change the paradigms of both software development and use. This panel will discuss the current status and challenges of Web services technologies.

VRML (Virtual Reality Modeling Language) is a programming language with promising effectiveness in describing cyberspace because of its real-time expressive power for 3D objects via the Internet. Concerning its 3D design, CAD tools can support static feature developments effectively, but currently, developments of dynamic features such as animation cannot be supported well, thus the tasks of debugging, testing, and behavior parameter adjustment become difficult. A program slicing technique is effective for such tasks. Thus we have applied it into VRML programs. In this paper, we studied VRML program structures and features, then proposed various kinds of dependency in VRML, and also proposed a VRML slicing approach based on such dependency. Finally we implemented several support tools.

An algorithm for the extraction of character strings from a color document is proposed. We first divide the full color image of a document into several representative color images. Then, character strings are nominated from each binary image by using multi-stage relaxation. However, the nominated strings are not always characters. Therefore, when all nominated strings of all images are superimposed, some strings overlap each other. So, we selected the appropriate strings from them using a likelihood of a character string and two binds of conflict resolution. Finally, we show the results of the experiments and discuss some problems of character string extraction from a color document.

An extraction algorithm for character strings is proposed. We first obtain a set of eight-connected components from a document image. For the components, we apply a relaxation method. The method makes mutual connections between components increase or decrease depending on the state of the neighboring components. While applying the relaxation method several times, the process proceeds from a local connection to a global connection, and finally character strings are extracted. We call this process multi stage relaxation. The advantages of this algorithm are that it does not need to nominate character components from an image beforehand, it is adaptive for character size and font, and it can also cope with a document which includes strings with various orientations. In our experiments we use a color image of a magazine cover and a monochromatic image of a graph. For the color image, the multi stage relaxation was executed for each binary image obtained by color segmentation. Lastly, we show the results of the experiments and discuss the effectiveness of our method.

We address the problem of automatic recognition of unconstrained handwritten text. Statistical methods, such as hidden Markov models (HMMs) have been used successfully for speech recognition and they have been applied to the problem of handwriting recognition as well. We discuss a general recognition system for large vocabulary, writer independent, unconstrained handwritten text. "Unconstrained" implies that the user may write in any style e.g. printed, cursive or in any combination of styles. This is more representative of typical handwritten text where one seldom encounters purely printed or purely cursive forms. Furthermore, a key characteristic of the system is that it performs recognition in real-time on 486 class PC platforms without the large amounts of memory required for traditional HMM based systems. We focus mainly on the writer independent task. Some initial writer dependent results are also reported. An error rate of 18.9% is achieved for a writer-independent 21,000 word vocabulary task in the absence of any language models.

The authors focus on strongly typed functional programming languages with polymorphic types, type inference facilities, and higher-order functions, i.e. functional programming languages whose type checkers infer types in programs including polymorphic types and higher-order functions types from the context before execution programs. They examine effective approaches to support type error corrections in these languages and two approaches to support them, following a general framework for debugging. Using these two approaches as case studies, systematic debugging processes based on the general framework for debugging are developed.>

We present a new grammatical formalism called Constraint Dependency Grammar (CDG) in which every grammatical rule is given as a constraint on word-to-word modifications. CDG parsing is formalized as a constraint satisfaction problem over a finite domain so that efficient constraint-propagation algorithms can be employed to reduce structural ambiguity without generating individual parse trees. The weak generative capacity and the computational complexity of CDG parsing are also discussed.