Location: Main Library- PK6420.S8I84 2014

Location: Main Library- PK6406.B375 2014

Ole Løseth Elvetun and Bjørn Fredrik Nielsen
Math. Comp. 93 (), 2811-2836.
Abstract, references and article information

Tara Sutaria celebrated her Parsi Roj birthday at home with her close friends.

Over the years, I've given more than 300 presentations on design. Most of these have been accompanied by a slide deck to illustrate my points and guide the narrative. But making the content in these decks work well with the Ask Luke conversational interface on this site has been challenging. So now I'm trying a new approach with AI vision models.

To avoid application specific formats (Keynote, PowerPoint), I've long been making my presentation slides available for download as PDF documents. These files usually consist of 100+ pages and often don't include a lot of text, leaning instead on visuals and charts to communicate information. To illustrate, here's of few of these slides from my Mind the Gap talk.

In an earlier article on how we built the Ask Luke conversational interface, I outlined the issues with extracting useful information from these documents. I wanted the content in these PDFs to be available when answering people's design questions in addition to the blog articles, videos and audio interviews that we were already using.

But even when we got text extraction from PDFs working well, running the process on any given PDF document would create many content embeddings of poor quality (like the one below). These content chunks would then end up influencing the answers we generated in less than helpful ways.

To prevent these from clogging up our limited context (how much content we can work with to create an answer) with useless results, we set up processes to remove low quality content chunks. While that improved things, the content in these presentations was no longer accessible to people asking questions on Ask Luke.

So we tried a different approach. Instead of extracting text from each page of a PDF presentation, we ran it through an AI vision model to create a detailed description of the content on the page. In the example below, the previous text extraction method (on the left) gets the content from the slide. The new vision model approach (on the right) though, does a much better job creating useful content for answering questions.

Here's another example illustrating the difference between the PDF text extraction method used before and the vision AI model currently in use. This time instead of a chart, we're generating a useful description of a diagram.

This change is now rolled out across all the PDFs the Ask Luke conversational interface can reference to answer design questions. Gone are useless content chunks and there's a lot more useful content immediately available.

Thanks to Yangguang Li for the dev help on this change.

The Parsi community is the transition zone between the monotheistic and polytheistic mythologies of West and South Asia

The First in the State of New Jersey

An interesting and frequent type of multi-word expression (MWE) is the headless MWE, for which there are no true internal syntactic dominance relations; examples include many named entities ("Wells Fargo") and dates ("July 5, 2020") as well as certain productive constructions ("blow for blow", "day after day"). Despite their special status and prevalence, current dependency-annotation schemes require treating such flat structures as if they had internal syntactic heads, and most current parsers handle them in the same fashion as headed constructions. Meanwhile, outside the context of parsing, taggers are typically used for identifying MWEs, but taggers might benefit from structural information. We empirically compare these two common strategies--parsing and tagging--for predicting flat MWEs. Additionally, we propose an efficient joint decoding algorithm that combines scores from both strategies. Experimental results on the MWE-Aware English Dependency Corpus and on six non-English dependency treebanks with frequent flat structures show that: (1) tagging is more accurate than parsing for identifying flat-structure MWEs, (2) our joint decoder reconciles the two different views and, for non-BERT features, leads to higher accuracies, and (3) most of the gains result from feature sharing between the parsers and taggers.

Systems and methods for generating a tuple of structured data files are described herein. In one example, a method includes detecting an expression that describes a structure of a structured image using a constructor. The method can also include using an inference-rule based search strategy to identify a hierarchical arrangement of bounding boxes in the structured image that match the expression. Furthermore, the method can include generating a first tuple of structured data files based on the identified hierarchical arrangement of bounding boxes in the structured image.

A method for performing relational analysis of parsed input is employed to create a visual map of knowledge information. A title, header or subject line for an input item of information is parsed into syntactical components of at least a subject component and any predicate component(s) relationally linked as topic and subtopics. A search of topics and subtopics is carried out for each parsed component. If a match is found, then the parsed component is taken as a chosen topic/subtopic label. If no match is found, then the parsed component is formatted as a new entry in the knowledge map. A translation function for translating topics and subtopics from an original language into one or more target languages is enabled by user request or indicated user preference for display on a generated visual map of knowledge information.

Methods and apparatus for parsing friendly and error resilient merge flag coding in video coding are provided. In some methods, in contrast to merging candidate list size dependent coding of the merge flag in the prior art, a merge flag is always encoded in the encoded bit stream for each inter-predicted prediction unit (PU) that is not encoded using skip mode. In some methods, in contrast to the prior art that allowed the merging candidate list to be empty, one or more zero motion vector merging candidates formatted according to the prediction type of the slice containing a PU are added to the merging candidate list if needed to ensure that the list is not empty and/or to ensure that the list contains a maximum number of merging candidates.

Described are methods, systems and computer readable media for parsing and compiling data system queries.

Aurélie Fischer, Dominique Picard.

Source: Electronic Journal of Statistics, Volume 14, Number 1, 1648--1689.

Abstract:
In this paper, we consider the estimation of a change-point for possibly high-dimensional data in a Gaussian model, using a maximum likelihood method. We are interested in how dimension reduction can affect the performance of the method. We provide an estimator of the change-point that has a minimax rate of convergence, up to a logarithmic factor. The minimax rate is in fact composed of a fast rate —dimension-invariant— and a slow rate —increasing with the dimension. Moreover, it is proved that considering the case of sparse data, with a Sobolev regularity, there is a bound on the separation of the regimes above which there exists an optimal choice of dimension reduction, leading to the fast rate of estimation. We propose an adaptive dimension reduction procedure based on Lepski’s method and show that the resulting estimator attains the fast rate of convergence. Our results are then illustrated by a simulation study. In particular, practical strategies are suggested to perform dimension reduction.

A deep neural network has relieved the burden of feature engineering by human experts, but comparable efforts are instead required to determine an effective architecture. On the other hands, as the size of a network has over-grown, a lot of resources are also invested to reduce its size. These problems can be addressed by sparsification of an over-complete model, which removes redundant parameters or connections by pruning them away after training or encouraging them to become zero during training. In general, however, these approaches are not fully differentiable and interrupt an end-to-end training process with the stochastic gradient descent in that they require either a parameter selection or a soft-thresholding step. In this paper, we propose a fully differentiable sparsification method for deep neural networks, which allows parameters to be exactly zero during training, and thus can learn the sparsified structure and the weights of networks simultaneously using the stochastic gradient descent. We apply the proposed method to various popular models in order to show its effectiveness.

Joseph Antonelli, Maitreyi Mazumdar, David Bellinger, David Christiani, Robert Wright, Brent Coull.

Source: The Annals of Applied Statistics, Volume 14, Number 1, 257--275.

Abstract:
Humans are routinely exposed to mixtures of chemical and other environmental factors, making the quantification of health effects associated with environmental mixtures a critical goal for establishing environmental policy sufficiently protective of human health. The quantification of the effects of exposure to an environmental mixture poses several statistical challenges. It is often the case that exposure to multiple pollutants interact with each other to affect an outcome. Further, the exposure-response relationship between an outcome and some exposures, such as some metals, can exhibit complex, nonlinear forms, since some exposures can be beneficial and detrimental at different ranges of exposure. To estimate the health effects of complex mixtures, we propose a flexible Bayesian approach that allows exposures to interact with each other and have nonlinear relationships with the outcome. We induce sparsity using multivariate spike and slab priors to determine which exposures are associated with the outcome and which exposures interact with each other. The proposed approach is interpretable, as we can use the posterior probabilities of inclusion into the model to identify pollutants that interact with each other. We utilize our approach to study the impact of exposure to metals on child neurodevelopment in Bangladesh and find a nonlinear, interactive relationship between arsenic and manganese.

Bo Ning, Seonghyun Jeong, Subhashis Ghosal.

Source: Bernoulli, Volume 26, Number 3, 2353--2382.

Abstract:
We study frequentist properties of a Bayesian high-dimensional multivariate linear regression model with correlated responses. The predictors are separated into many groups and the group structure is pre-determined. Two features of the model are unique: (i) group sparsity is imposed on the predictors; (ii) the covariance matrix is unknown and its dimensions can also be high. We choose a product of independent spike-and-slab priors on the regression coefficients and a new prior on the covariance matrix based on its eigendecomposition. Each spike-and-slab prior is a mixture of a point mass at zero and a multivariate density involving the $ell_{2,1}$-norm. We first obtain the posterior contraction rate, the bounds on the effective dimension of the model with high posterior probabilities. We then show that the multivariate regression coefficients can be recovered under certain compatibility conditions. Finally, we quantify the uncertainty for the regression coefficients with frequentist validity through a Bernstein–von Mises type theorem. The result leads to selection consistency for the Bayesian method. We derive the posterior contraction rate using the general theory by constructing a suitable test from the first principle using moment bounds for certain likelihood ratios. This leads to posterior concentration around the truth with respect to the average Rényi divergence of order $1/2$. This technique of obtaining the required tests for posterior contraction rate could be useful in many other problems.

launchd on macOS and iOS suffer from a memory corruption issue due to a lack of bounds checking when parsing XPC messages.

This article presents a cross-platform test harness written in Python that assists the user in searching for vulnerabilities in web browsers, specifically by fuzzing their font parsing functionality. The tool automates the delivery of test cases (font files in this context) into a web browser. The creation of a corpus of mutated TTF font files suitable for use in fuzzing is also covered.

The Iranian Foreign Minister mentioned that the Parsis are in fact Zoroastrians who migrated to India centuries ago.

A high court order has inadvertently turned the Dadar Parsi Colony into the Dadar Parking colony. Miffed with all kinds of taxis parking in its premises -after they were not allowed to park under the Khodadad flyover - and causing a 'great nuisance' to residents, the colony's residents wrote to the local police and the BMC in November, seeking an end to the issue. They have got no respite yet.

The Bombay High Court disallowed parking underneath flyovers in February 2016 because of security concerns. Because of that, cabbies have shifted from parking underneath the Khodadad Circle flyover to the Colony.

Cabbies have shifted from parking under the flyover to Dadar Parsi Colony

This has angered residents of the iconic community housing enclave in Dadar East, who first sent an anguished letter about their grievances to the Sion Police Station. The letter cited that private taxi aggregators, black and yellow cabs, and other cabs are parking in the colony, causing a, "great nuisance" to residents, especially those of Katrak, Firdoshi, Khareghat and Mancherji Joshi roads within the colony. Their letter went on to say that cabs are being parked haphazardly, blocking traffic within the colony and not giving its residents any space to park their cars.

Residents Darayus (who goes by his first name), Zarine Engineer, Dilly Dalal, Anjali Cooper, Zinobia Schroff, Sunita Davar and Nikhil Desai said, "The problem started after parking under flyovers was stopped by a High Court order. We appreciate this security measure, yet a majority the cabs that used to be parked under the Khodadad flyover have now made Dadar Parsi Colony their home. This has become a free parking spot."

Garden pardon
When mid-day visited the flyover, underneath which cabs used to be parked, it was clear of vehicles and shuttered from one side. The cab booking office though was functional.

A staffer sat at the ticket window for Mumbai-Pune cabs and said sarcastically, "Bageecha bana rahe hain, pareshaan kar diya." (A garden is being made here; we have been troubled because of no parking).

Said Dadar colony resident Dilly Dalal, "We have become a soft target because the community is so peaceful; we do not stoop to rowdy protests or resort to hooliganism." Instead, these Parsis believe in the power of the pen. Another letter was sent in November, to the BMC's assistant engineer of the waterworks department (F/North Ward), which encapsulated residents' arguments about cab parking bringing in social ills.

This letter read, "There is an illegal racket on at Katrak Road Junction. A flower merchant draws water illegally from a toilet meant for shopkeepers and then the cabs are cleaned." Residents pointed out the huge amount of cab washing makes the roads slippery, and has turned the place into "an illegal taxi stand." They stated that cab parking has a domino effect, "it leads to problems like drinking and other anti-social behaviour."

Residents said angrily that, "There is double and triple parking because of these cabs and we have been rendered helpless against this." Desai added, "Despite complaints, the system is not responding to the problem."

Locals were also in no mood to hum 'dil garden, garden ho gaya' when told that a beautiful garden is going to come up under the flyover, which they could access, pointing to at least 14 gardens within the colony itself. Cooper said, "Authorities should have found an alternate and viable parking space before shutting the parking underneath the flyover."

No solution
Amey Gole, the corporator for Dadar Parsi Colony, said he has had several meetings with residents, but nothing has come out of them. Gole added, "I am, in fact, in-charge of the garden coming up underneath the flyover. The design is almost complete; it looks like the streets of Paris."

Gole added, "Having said that, I do know the problems being caused by parked cabs, and have spoken to the police about these. The police though had a negative attitude and said they could not move the cabs out of the colony. Cops have to ensure the cabs move out, not the BMC. We do have other parking spaces in the area. All of this is extremely difficult as I have been repeatedly told there is no solution," finished Gole.

Also read: Mumbai: Construction noise leads to brawl at Dadar Parsi Colony

Catch up on all the latest Mumbai news, crime news, current affairs, and also a complete guide on Mumbai from food to things to do and events across the city here. Also download the new mid-day Android and iOS apps to get latest updates

I am involved in several free software projects, including one or two where I maintain the website. For one of those projects, we currently are updating the website. Ours is probably similar to other free software projects. We use a hosting service for several key services, including news, but we run our website on a web server that we own. In our case, we run most of our project on SourceForge and run the website on a third-party service, so the news and website are on different systems.

Not surprisingly, our project uses an RSS feed to pull news items from SourceForge to display on the project website.

complete article

Gupta, Somanātha

Use this link and code JAVAJAB to get 20% off your registration for FluentConf 2013! Panel David Herman (twitter blog Effective JavaScript) Ariya Hidayat (twitter github blog) Tim Caswell (twitter github howtonode.org) Jamison Dance (twitter github blog) Joe Eames (twitter github blog) Merrick Christensen (twitter github) Charles Max Wood (twitter github Teach Me To Code Rails Ramp Up) Discussion 00:48 - David Herman and Ariya Hidayat Introduction 044 JSJ Book Club: Effective JavaScript with David Herman 023 JSJ Phantom.js with Ariya Hidayat 01:54 - Parsing JavaScript and ASTs and Language Grammars 04:44 - Semantics 06:08 - Abstract Syntax Tree (AST) Esprima: Parser SpiderMonkey 10:37 - Lexer 12:16 - Writing your own language creationix / jack The C Programming Language 17:41 - Parser Generators JavaScriptCore 21:04 - Evolving a Syntax Automatic Semicolon Insertion Post correspondence problem Halting problem 28:05 - Language Design The Rust Programming Language 30:35 - Grammar Regular Expressions (Regex) Backus–Naur Form (BNF) Recursion How to Design Programs (HTDP) 38:00 - Recursive Descent Parsers 42:48 - Benefits of knowing language internals and syntax Apache Lucene - Apache Lucene Core LPeg - Parsing Expression Grammars For Lua 48:48 - Abstract Syntax Tree (AST) Picks Mass Effect 3 (Joe) A Beginner's Guide to Irrational Behavior | Coursera (Joe) Go write a programming language to learn one (Tim) Thumbs and Ammo (Jamison) ISM by Savant (Jamison) Vimcasts (Jamison) The iPhreaks Show (Chuck) Mozy (Chuck) Tech & Go Bright Pink Micro USB Cable (David) asm.js (David) Beyond Office Politics: The Hidden Story of Power, Affiliation & Achievement in the Workplace by Linda Sommer (Ariya) gotwarlost / istanbul (Ariya) Next Week Web Developer Skills Transcript JAMISON:  I am Linus Torvalds and I pronounce Linux, Linix. [Hosting and bandwidth provided by the Blue Box Group. Check them out at Bluebox.net.] [This episode is sponsored by Component One, makers of Wijmo. If you need stunning UI elements or awesome graphs and charts, then go to Wijmo.com and check them out.] CHUCK:  Hey everybody and welcome to Episode 54 of the JavaScript Jabber Show. This week on our panel, we have Tim Caswell. TIM:  Hello. CHUCK:  Jamison Dance. JAMISON:  Hi guys. CHUCK:  Joe Eames. JOE:  Hey there. CHUCK:  Merrick Christensen. MERRICK:  Hey guys, what’s up? CHUCK:  I’m Charles Max Wood from DevChat.tv. And we have two special guests this week. We have Dave Herman. DAVID:  Hey there. CHUCK:  Ariya Hidayat. ARIYA:  Hello everyone. CHUCK:  And these guys are so smart that we brought them back. So, if you’re interested, we’ll put links to the episodes that they were on. David was on when we talked about his book ‘Essential JavaScript’ and Ariya was on when we talked about PhantomJS. JAMISON:  Effective JavaScript. CHUCK:  Effective? What did I say? MERRICK:  Essential. CHUCK:  Essential? Well, it’s an essential book on Effective JavaScript. How’s that? [Laughter] MERRICK:  Good save. DAVID:  At least, you didn’t say Defective JavaScript. [Laughter] CHUCK:  No, that’s what I write. I’m really good at writing defective JavaScript. ARIYA:  Actually, there’s a book about Essential on Defective JavaScript. CHUCK:  I also want to announce really quickly that Fluent Conf has given us a discount code. So, if you want to get 20% off on your registration for Fluent Conf, just enter JAVAJAB and you’ll get 20% off when you register for Fluent Conf. Alright. Well, let’s get started. This is going to be a really, really interesting topic and it’s something that I’ve wanted to know more about for a long time. And I just haven’t delved as deeply into it as I would like to. And that is,

Rotch Library - PJ5089.E4713 2016