Kevin H. Lai - Academia.edu (original) (raw)

Papers by Kevin H. Lai

Proceedings of the AAAI Conference on Artificial Intelligence

Recognizing possibly thousands of objects is a crucial capability for an autonomous agent to unde... more Recognizing possibly thousands of objects is a crucial capability for an autonomous agent to understand and interact with everyday environments. Practical object recognition comes in multiple forms: Is this a coffee mug (category recognition). Is this Alice's coffee mug? (instance recognition). Is the mug with the handle facing left or right? (pose recognition). We present a scalable framework, Object-Pose Tree, which efficiently organizes data into a semantically structured tree. The tree structure enables both scalable training and testing, allowing us to solve recognition over thousands of object poses in near real-time. Moreover, by simultaneously optimizing all three tasks, our approach outperforms standard nearest neighbor and 1-vs-all classifications, with large improvements on pose recognition. We evaluate the proposed technique on a dataset of 300 household objects collected using a Kinect-style 3D camera. Experiments demonstrate that our system achieves robust and effi...

AIP Conference Proceedings, 2005

We propose a novel stress metrology technique for measurement of local stress tensor components. ... more We propose a novel stress metrology technique for measurement of local stress tensor components. Metrology is based on fiber-optic low coherence interferometry and can be applied to study stress not only in semiconductor wafers but in other applications such as displays, solar cells, modern windows.

Proceedings of the 6th annual international conference on Mobile computing and networking - MobiCom '00, 2000

This paper describes two techniques that improve throughput in an ad hoc network in the presence ... more This paper describes two techniques that improve throughput in an ad hoc network in the presence of nodes that agree to forward packets but fail to do so. To mitigate this problem, we propose categorizing nodes based upon their dynamically measured behavior. We use a watchdog that identifies misbehaving nodes and a patl~rater that helps routing protocols avoid these nodes. Through simulation we evaluate watchdog and pathrater using packet throughput, percentage of overhead (routing) transmissions, and the accuracy of misbehaving node detection. When used together in a network with moderate mobility, the two techniques increase throughput by 17% in the presence of 40% misbehaving nodes, while increasing the percentage of overhead transmissions from the standard routing protocol's 9% to 17%. During extreme mobility, watchdog and pathrater can increase network throughput by 27%, while increasing the overhead transmissions from the standard routing protocol's 12% to 24%.

SPIE Proceedings, 2005

Low coherence optical interferometry has been proven to be an effective tool for characterization... more Low coherence optical interferometry has been proven to be an effective tool for characterization of thin and ultra-thin, transparent and non-transparent semiconductor Si and compound wafers, and MEMs structures for ex-situ and in-situ applications. We demonstrate that use of synchronously operating probes significantly reduces vibration noise observed in the system. We demonstrate that application of synchronized improves reproducibility of measurement

ACM SIGecom Exchanges, 2005

Researchers have long proposed using economic approaches to resource allocation in computer syste... more Researchers have long proposed using economic approaches to resource allocation in computer systems. However, few of these proposals became operational, let alone commercial. Questions persist about the economic approach regarding its assumptions, value, applicability, and relevance to system design. The goal of this paper is to answer these questions. We find that market-based resource allocation is useful, and more importantly, that mechanism design and system design should be integrated to produce systems that are both economically and computationally efficient.

2014 IEEE International Conference on Robotics and Automation (ICRA), 2014

This paper presents an approach for labeling objects in 3D scenes. We introduce HMP3D, a hierarch... more This paper presents an approach for labeling objects in 3D scenes. We introduce HMP3D, a hierarchical sparse coding technique for learning features from 3D point cloud data. HMP3D classifiers are trained using a synthetic dataset of virtual scenes generated using CAD models from an online database. Our scene labeling system combines features learned from raw RGB-D images and 3D point clouds directly, without any hand-designed features, to assign an object label to every 3D point in the scene. Experiments on the RGB-D Scenes Dataset v.2 demonstrate that the proposed approach can be used to label indoor scenes containing both small tabletop objects and large furniture pieces.

Consumer Depth Cameras for Computer Vision, 2013

ABSTRACT Over the last decade, the availability of public image repositories and recognition benc... more ABSTRACT Over the last decade, the availability of public image repositories and recognition benchmarks has enabled rapid progress in visual object category and instance detection. Today we are witnessing the birth of a new generation of sensing technologies capable of providing high quality synchronized videos of both color and depth, the RGB-D (Kinect-style) camera. With its advanced sensing capabilities and the potential for mass adoption, this technology represents an opportunity to dramatically increase robotic object recognition, manipulation, navigation, and interaction capabilities. We introduce a large-scale, hierarchical multi-view object dataset collected using an RGB-D camera. The dataset consists of two parts: The RGB-D Object Dataset containing views of 300 objects organized into 51 categories, and the RGB-D Scenes Dataset containing 8 video sequences of office and kitchen environments. The dataset has been made publicly available to the research community so as to enable rapid progress based on this promising technology. We describe the dataset collection procedure and present techniques for RGB-D object recognition and detection of objects in scenes recorded using RGB-D videos, demonstrating that combining color and depth information substantially improves quality of results.

2011 IEEE International Conference on Robotics and Automation, 2011

Over the last decade, the availability of public image repositories and recognition benchmarks ha... more Over the last decade, the availability of public image repositories and recognition benchmarks has enabled rapid progress in visual object category and instance detection. Today we are witnessing the birth of a new generation of sensing technologies capable of providing high quality synchronized videos of both color and depth, the RGB-D (Kinectstyle) camera. With its advanced sensing capabilities and the potential for mass adoption, this technology represents an opportunity to dramatically increase robotic object recognition, manipulation, navigation, and interaction capabilities. In this paper, we introduce a large-scale, hierarchical multi-view object dataset collected using an RGB-D camera. The dataset contains 300 objects organized into 51 categories and has been made publicly available to the research community so as to enable rapid progress based on this promising technology. This paper describes the dataset collection procedure and introduces techniques for RGB-D based object recognition and detection, demonstrating that combining color and depth information substantially improves quality of results.

Emergency Medicine Journal, 2014

Objective To determine whether a narrower gauge needle used in ABG sampling is associated with lo... more Objective To determine whether a narrower gauge needle used in ABG sampling is associated with lower pain scores and complication rates without increasing the level of difficulty of the procedure. Methods We performed a prospective single-blinded randomised control study of patients from a tertiary-level emergency department in Sydney who required an ABG analysis over the period of June 2010-July 2012. Patients were randomised to either a 23G or 25G needle and the primary outcome that included pain experienced by these patient were recorded as pain scores on a 10 cm hatched visual analogue scale. The difficulty scores and complications were also noted from the operator. Results Data for 119 consenting eligible patients were included in the analysis. 63 patients were allocated to the 23G needle group and 56 to the 25G needle group. The mean pain score was 3.5 (SD=2.7) for the 23G group and 3.4 (SD=2.7) for the 25G group with a mean difference between the pain scores of 0.1 (95% CI −0.9 to 1.1, p=0.83). The 23G and 25G mean difficulty score was 3.4 (SD=2.6) and 4.3 (SD=2.4), respectively, with a mean difference of 0.9 (95% CI −0.03 to 1.7, p=0.06). 21.6% of patient in the 23G needle group experienced some complication with regard to the sampling in the form of haematoma, tenderness or paraesthesia in comparison to 5.4% of patients in the 25G needle group (p=0.03). Conclusions There was no significant difference in pain scores experienced by patients undertaking ABG sampling with either a 23G or 25G needle. Trial registration number ACTRN12609000957291.

Lecture Notes in Computer Science, 1998

Abstract. This paper presents results from an eight-day network packet trace of MosquitoNet. Mosq... more Abstract. This paper presents results from an eight-day network packet trace of MosquitoNet. MosquitoNet allows users of laptop computers to switch seamlessiy between a metropohtan-area wireless network and a wired network (10 Mbit/s Ethernet) available in ...

Proceedings of the 8th iternational conference on Computer supported collaborative learning - CSCL'07, 2007

This paper explores the potential of networked devices to support classroom problem solving in sm... more This paper explores the potential of networked devices to support classroom problem solving in small groups. We articulate two principles for designing networked collaborative activities: that they should 1) balance the group's collective engagement of shared objects with opportunities for individual student manipulation of those objects and 2) coordinate networked interactions among student-controlled objects with mathematically meaningful relationships. To illustrate these principles, we present a scenario for small-group collaboration involving classroom device networks.

2011 IEEE International Conference on Robotics and Automation, 2011

In this work we address joint object category and instance recognition in the context of RGB-D (d... more In this work we address joint object category and instance recognition in the context of RGB-D (depth) cameras. Motivated by local distance learning, where a novel view of an object is compared to individual views of previously seen objects, we define a view-to-object distance where a novel view is compared simultaneously to all views of a previous object. This novel distance is based on a weighted combination of feature differences between views. We show, through jointly learning perview weights, that this measure leads to superior classification performance on object category and instance recognition. More importantly, the proposed distance allows us to find a sparse solution via Group-Lasso regularization, where a small subset of representative views of an object is identified and used, with the rest discarded. This significantly reduces computational cost without compromising recognition accuracy. We evaluate the proposed technique, Instance Distance Learning (IDL), on the RGB-D Object Dataset, which consists of 300 object instances in 51 everyday categories and about 250,000 views of objects with both RGB color and depth. We empirically compare IDL to several alternative state-of-the-art approaches and also validate the use of visual and shape cues and their combination.

2012 IEEE International Conference on Robotics and Automation, 2012

We propose a view-based approach for labeling objects in 3D scenes reconstructed from RGB-D (colo... more We propose a view-based approach for labeling objects in 3D scenes reconstructed from RGB-D (color+depth) videos. We utilize sliding window detectors trained from object views to assign class probabilities to pixels in every RGB-D frame. These probabilities are projected into the reconstructed 3D scene and integrated using a voxel representation. We perform efficient inference on a Markov Random Field over the voxels, combining cues from view-based detection and 3D shape, to label the scene. Our detection-based approach produces accurate scene labeling on the RGB-D Scenes Dataset and improves the robustness of object detection.

CVPR 2011, 2011

Kernel descriptors [1] provide a unified way to generate rich visual feature sets by turning pixe... more Kernel descriptors [1] provide a unified way to generate rich visual feature sets by turning pixel attributes into patch-level features, and yield impressive results on many object recognition tasks. However, best results with kernel descriptors are achieved using efficient match kernels in conjunction with nonlinear SVMs, which makes it impractical for large-scale problems. In this paper, we propose hierarchical kernel descriptors that apply kernel descriptors recursively to form image-level features and thus provide a conceptually simple and consistent way to generate imagelevel features from pixel attributes. More importantly, hierarchical kernel descriptors allow linear SVMs to yield stateof-the-art accuracy while being scalable to large datasets. They can also be naturally extended to extract features over depth images. We evaluate hierarchical kernel descriptors both on the CIFAR10 dataset and the new RGB-D Object Dataset consisting of segmented RGB and depth images of 300 everyday objects.

Cornea, 2014

The aim of this study was to describe clinical, imaging, molecular genetic, histopathologic, immu... more The aim of this study was to describe clinical, imaging, molecular genetic, histopathologic, immunohistochemical, and ultrastructural characteristics of coexistent amyloid and spheroidal degeneration-type deposits in a family with histidine-626-arginine transforming growth factor beta-induced (H626R TGFBI) variant lattice dystrophy. This is a retrospective clinical-pathological and genetic analysis of one family with H626R variant lattice dystrophy. Pedigree analysis showed an autosomal dominant inheritance pattern of the disease. Examination of 3 affected family members revealed asymmetric, thick, branching lattice-like deposits associated with corneal haze. Sequencing of the TGFBI gene revealed a high-penetrance disease-causing sequence variation (H626R CAT>CGT heterozygous). Optical coherence tomography demonstrated fusiform, poorly demarcated hyperechoic stromal deposits with focal hypoechoic central regions. Histology of the corneal discs from 2 affected family members showe...

PLoS Biology, 2004

We have generated several hundred lines of zebrafish (Danio rerio), each heterozygous for a reces... more We have generated several hundred lines of zebrafish (Danio rerio), each heterozygous for a recessive embryonic lethal mutation. Since many tumor suppressor genes are recessive lethals, we screened our colony for lines that display early mortality and/or gross evidence of tumors. We identified 12 lines with elevated cancer incidence. Fish from these lines develop malignant peripheral nerve sheath tumors, and in some cases also other tumor types, with moderate to very high frequencies. Surprisingly, 11 of the 12 lines were each heterozygous for a mutation in a different ribosomal protein (RP) gene, while one line was heterozygous for a mutation in a zebrafish paralog of the human and mouse tumor suppressor gene, neurofibromatosis type 2. Our findings suggest that many RP genes may act as haploinsufficient tumor suppressors in fish. Many RP genes might also be cancer genes in humans, where their role in tumorigenesis could easily have escaped detection up to now.

ACM SIGMOBILE Mobile Computing and Communications Review, 1999

People are the outsiders in the current communications revolution. Computer hosts, pagers, and te... more People are the outsiders in the current communications revolution. Computer hosts, pagers, and telephones are the addressable entities throughout the Internet and telephony systems. Human beings, however, still need application-specific tricks to be identified, like email addresses, telephone numbers, and ICQ IDs. The key challenge today is to find people and communicate with them personally, as opposed to communicating merely with their possibly inaccessible machines---cell phones that are turned off or PCs on faraway desktops. We introduce the Mobile People Architecture which aims to put the person, rather than the devices that the person uses, at the endpoints of a communication session. We describe a prototype that performs person-level routing; the prototype allows people to receive communication regardless of the network, device, or application they use, while maintaining their privacy.

Proceedings of the AAAI Conference on Artificial Intelligence

Recognizing possibly thousands of objects is a crucial capability for an autonomous agent to unde... more Recognizing possibly thousands of objects is a crucial capability for an autonomous agent to understand and interact with everyday environments. Practical object recognition comes in multiple forms: Is this a coffee mug (category recognition). Is this Alice's coffee mug? (instance recognition). Is the mug with the handle facing left or right? (pose recognition). We present a scalable framework, Object-Pose Tree, which efficiently organizes data into a semantically structured tree. The tree structure enables both scalable training and testing, allowing us to solve recognition over thousands of object poses in near real-time. Moreover, by simultaneously optimizing all three tasks, our approach outperforms standard nearest neighbor and 1-vs-all classifications, with large improvements on pose recognition. We evaluate the proposed technique on a dataset of 300 household objects collected using a Kinect-style 3D camera. Experiments demonstrate that our system achieves robust and effi...

AIP Conference Proceedings, 2005

We propose a novel stress metrology technique for measurement of local stress tensor components. ... more We propose a novel stress metrology technique for measurement of local stress tensor components. Metrology is based on fiber-optic low coherence interferometry and can be applied to study stress not only in semiconductor wafers but in other applications such as displays, solar cells, modern windows.

Proceedings of the 6th annual international conference on Mobile computing and networking - MobiCom '00, 2000

This paper describes two techniques that improve throughput in an ad hoc network in the presence ... more This paper describes two techniques that improve throughput in an ad hoc network in the presence of nodes that agree to forward packets but fail to do so. To mitigate this problem, we propose categorizing nodes based upon their dynamically measured behavior. We use a watchdog that identifies misbehaving nodes and a patl~rater that helps routing protocols avoid these nodes. Through simulation we evaluate watchdog and pathrater using packet throughput, percentage of overhead (routing) transmissions, and the accuracy of misbehaving node detection. When used together in a network with moderate mobility, the two techniques increase throughput by 17% in the presence of 40% misbehaving nodes, while increasing the percentage of overhead transmissions from the standard routing protocol's 9% to 17%. During extreme mobility, watchdog and pathrater can increase network throughput by 27%, while increasing the overhead transmissions from the standard routing protocol's 12% to 24%.

SPIE Proceedings, 2005

Low coherence optical interferometry has been proven to be an effective tool for characterization... more Low coherence optical interferometry has been proven to be an effective tool for characterization of thin and ultra-thin, transparent and non-transparent semiconductor Si and compound wafers, and MEMs structures for ex-situ and in-situ applications. We demonstrate that use of synchronously operating probes significantly reduces vibration noise observed in the system. We demonstrate that application of synchronized improves reproducibility of measurement

ACM SIGecom Exchanges, 2005

Researchers have long proposed using economic approaches to resource allocation in computer syste... more Researchers have long proposed using economic approaches to resource allocation in computer systems. However, few of these proposals became operational, let alone commercial. Questions persist about the economic approach regarding its assumptions, value, applicability, and relevance to system design. The goal of this paper is to answer these questions. We find that market-based resource allocation is useful, and more importantly, that mechanism design and system design should be integrated to produce systems that are both economically and computationally efficient.

2014 IEEE International Conference on Robotics and Automation (ICRA), 2014

This paper presents an approach for labeling objects in 3D scenes. We introduce HMP3D, a hierarch... more This paper presents an approach for labeling objects in 3D scenes. We introduce HMP3D, a hierarchical sparse coding technique for learning features from 3D point cloud data. HMP3D classifiers are trained using a synthetic dataset of virtual scenes generated using CAD models from an online database. Our scene labeling system combines features learned from raw RGB-D images and 3D point clouds directly, without any hand-designed features, to assign an object label to every 3D point in the scene. Experiments on the RGB-D Scenes Dataset v.2 demonstrate that the proposed approach can be used to label indoor scenes containing both small tabletop objects and large furniture pieces.

Consumer Depth Cameras for Computer Vision, 2013

ABSTRACT Over the last decade, the availability of public image repositories and recognition benc... more ABSTRACT Over the last decade, the availability of public image repositories and recognition benchmarks has enabled rapid progress in visual object category and instance detection. Today we are witnessing the birth of a new generation of sensing technologies capable of providing high quality synchronized videos of both color and depth, the RGB-D (Kinect-style) camera. With its advanced sensing capabilities and the potential for mass adoption, this technology represents an opportunity to dramatically increase robotic object recognition, manipulation, navigation, and interaction capabilities. We introduce a large-scale, hierarchical multi-view object dataset collected using an RGB-D camera. The dataset consists of two parts: The RGB-D Object Dataset containing views of 300 objects organized into 51 categories, and the RGB-D Scenes Dataset containing 8 video sequences of office and kitchen environments. The dataset has been made publicly available to the research community so as to enable rapid progress based on this promising technology. We describe the dataset collection procedure and present techniques for RGB-D object recognition and detection of objects in scenes recorded using RGB-D videos, demonstrating that combining color and depth information substantially improves quality of results.

2011 IEEE International Conference on Robotics and Automation, 2011

Over the last decade, the availability of public image repositories and recognition benchmarks ha... more Over the last decade, the availability of public image repositories and recognition benchmarks has enabled rapid progress in visual object category and instance detection. Today we are witnessing the birth of a new generation of sensing technologies capable of providing high quality synchronized videos of both color and depth, the RGB-D (Kinectstyle) camera. With its advanced sensing capabilities and the potential for mass adoption, this technology represents an opportunity to dramatically increase robotic object recognition, manipulation, navigation, and interaction capabilities. In this paper, we introduce a large-scale, hierarchical multi-view object dataset collected using an RGB-D camera. The dataset contains 300 objects organized into 51 categories and has been made publicly available to the research community so as to enable rapid progress based on this promising technology. This paper describes the dataset collection procedure and introduces techniques for RGB-D based object recognition and detection, demonstrating that combining color and depth information substantially improves quality of results.

Emergency Medicine Journal, 2014

Objective To determine whether a narrower gauge needle used in ABG sampling is associated with lo... more Objective To determine whether a narrower gauge needle used in ABG sampling is associated with lower pain scores and complication rates without increasing the level of difficulty of the procedure. Methods We performed a prospective single-blinded randomised control study of patients from a tertiary-level emergency department in Sydney who required an ABG analysis over the period of June 2010-July 2012. Patients were randomised to either a 23G or 25G needle and the primary outcome that included pain experienced by these patient were recorded as pain scores on a 10 cm hatched visual analogue scale. The difficulty scores and complications were also noted from the operator. Results Data for 119 consenting eligible patients were included in the analysis. 63 patients were allocated to the 23G needle group and 56 to the 25G needle group. The mean pain score was 3.5 (SD=2.7) for the 23G group and 3.4 (SD=2.7) for the 25G group with a mean difference between the pain scores of 0.1 (95% CI −0.9 to 1.1, p=0.83). The 23G and 25G mean difficulty score was 3.4 (SD=2.6) and 4.3 (SD=2.4), respectively, with a mean difference of 0.9 (95% CI −0.03 to 1.7, p=0.06). 21.6% of patient in the 23G needle group experienced some complication with regard to the sampling in the form of haematoma, tenderness or paraesthesia in comparison to 5.4% of patients in the 25G needle group (p=0.03). Conclusions There was no significant difference in pain scores experienced by patients undertaking ABG sampling with either a 23G or 25G needle. Trial registration number ACTRN12609000957291.

Lecture Notes in Computer Science, 1998

Abstract. This paper presents results from an eight-day network packet trace of MosquitoNet. Mosq... more Abstract. This paper presents results from an eight-day network packet trace of MosquitoNet. MosquitoNet allows users of laptop computers to switch seamlessiy between a metropohtan-area wireless network and a wired network (10 Mbit/s Ethernet) available in ...

Proceedings of the 8th iternational conference on Computer supported collaborative learning - CSCL'07, 2007

This paper explores the potential of networked devices to support classroom problem solving in sm... more This paper explores the potential of networked devices to support classroom problem solving in small groups. We articulate two principles for designing networked collaborative activities: that they should 1) balance the group's collective engagement of shared objects with opportunities for individual student manipulation of those objects and 2) coordinate networked interactions among student-controlled objects with mathematically meaningful relationships. To illustrate these principles, we present a scenario for small-group collaboration involving classroom device networks.

2011 IEEE International Conference on Robotics and Automation, 2011

In this work we address joint object category and instance recognition in the context of RGB-D (d... more In this work we address joint object category and instance recognition in the context of RGB-D (depth) cameras. Motivated by local distance learning, where a novel view of an object is compared to individual views of previously seen objects, we define a view-to-object distance where a novel view is compared simultaneously to all views of a previous object. This novel distance is based on a weighted combination of feature differences between views. We show, through jointly learning perview weights, that this measure leads to superior classification performance on object category and instance recognition. More importantly, the proposed distance allows us to find a sparse solution via Group-Lasso regularization, where a small subset of representative views of an object is identified and used, with the rest discarded. This significantly reduces computational cost without compromising recognition accuracy. We evaluate the proposed technique, Instance Distance Learning (IDL), on the RGB-D Object Dataset, which consists of 300 object instances in 51 everyday categories and about 250,000 views of objects with both RGB color and depth. We empirically compare IDL to several alternative state-of-the-art approaches and also validate the use of visual and shape cues and their combination.

2012 IEEE International Conference on Robotics and Automation, 2012

We propose a view-based approach for labeling objects in 3D scenes reconstructed from RGB-D (colo... more We propose a view-based approach for labeling objects in 3D scenes reconstructed from RGB-D (color+depth) videos. We utilize sliding window detectors trained from object views to assign class probabilities to pixels in every RGB-D frame. These probabilities are projected into the reconstructed 3D scene and integrated using a voxel representation. We perform efficient inference on a Markov Random Field over the voxels, combining cues from view-based detection and 3D shape, to label the scene. Our detection-based approach produces accurate scene labeling on the RGB-D Scenes Dataset and improves the robustness of object detection.

CVPR 2011, 2011

Kernel descriptors [1] provide a unified way to generate rich visual feature sets by turning pixe... more Kernel descriptors [1] provide a unified way to generate rich visual feature sets by turning pixel attributes into patch-level features, and yield impressive results on many object recognition tasks. However, best results with kernel descriptors are achieved using efficient match kernels in conjunction with nonlinear SVMs, which makes it impractical for large-scale problems. In this paper, we propose hierarchical kernel descriptors that apply kernel descriptors recursively to form image-level features and thus provide a conceptually simple and consistent way to generate imagelevel features from pixel attributes. More importantly, hierarchical kernel descriptors allow linear SVMs to yield stateof-the-art accuracy while being scalable to large datasets. They can also be naturally extended to extract features over depth images. We evaluate hierarchical kernel descriptors both on the CIFAR10 dataset and the new RGB-D Object Dataset consisting of segmented RGB and depth images of 300 everyday objects.

Cornea, 2014

The aim of this study was to describe clinical, imaging, molecular genetic, histopathologic, immu... more The aim of this study was to describe clinical, imaging, molecular genetic, histopathologic, immunohistochemical, and ultrastructural characteristics of coexistent amyloid and spheroidal degeneration-type deposits in a family with histidine-626-arginine transforming growth factor beta-induced (H626R TGFBI) variant lattice dystrophy. This is a retrospective clinical-pathological and genetic analysis of one family with H626R variant lattice dystrophy. Pedigree analysis showed an autosomal dominant inheritance pattern of the disease. Examination of 3 affected family members revealed asymmetric, thick, branching lattice-like deposits associated with corneal haze. Sequencing of the TGFBI gene revealed a high-penetrance disease-causing sequence variation (H626R CAT>CGT heterozygous). Optical coherence tomography demonstrated fusiform, poorly demarcated hyperechoic stromal deposits with focal hypoechoic central regions. Histology of the corneal discs from 2 affected family members showe...

PLoS Biology, 2004

We have generated several hundred lines of zebrafish (Danio rerio), each heterozygous for a reces... more We have generated several hundred lines of zebrafish (Danio rerio), each heterozygous for a recessive embryonic lethal mutation. Since many tumor suppressor genes are recessive lethals, we screened our colony for lines that display early mortality and/or gross evidence of tumors. We identified 12 lines with elevated cancer incidence. Fish from these lines develop malignant peripheral nerve sheath tumors, and in some cases also other tumor types, with moderate to very high frequencies. Surprisingly, 11 of the 12 lines were each heterozygous for a mutation in a different ribosomal protein (RP) gene, while one line was heterozygous for a mutation in a zebrafish paralog of the human and mouse tumor suppressor gene, neurofibromatosis type 2. Our findings suggest that many RP genes may act as haploinsufficient tumor suppressors in fish. Many RP genes might also be cancer genes in humans, where their role in tumorigenesis could easily have escaped detection up to now.

ACM SIGMOBILE Mobile Computing and Communications Review, 1999

People are the outsiders in the current communications revolution. Computer hosts, pagers, and te... more People are the outsiders in the current communications revolution. Computer hosts, pagers, and telephones are the addressable entities throughout the Internet and telephony systems. Human beings, however, still need application-specific tricks to be identified, like email addresses, telephone numbers, and ICQ IDs. The key challenge today is to find people and communicate with them personally, as opposed to communicating merely with their possibly inaccessible machines---cell phones that are turned off or PCs on faraway desktops. We introduce the Mobile People Architecture which aims to put the person, rather than the devices that the person uses, at the endpoints of a communication session. We describe a prototype that performs person-level routing; the prototype allows people to receive communication regardless of the network, device, or application they use, while maintaining their privacy.