Information Rules is a 1999 book by Carl Shapiro and Hal Varian applying traditional economic theories to modern information-based technologies. The book examines commercial strategies appropriate to companies that deal in information, given the high "first copy" and low "subsequent copy" costs of information commodities, such as music CDs or original texts. == Content == The book examines competing standards, and how a company might influence widespread consumer acceptance of one over another, such as VHS versus Betamax, or HD DVD versus Blu-ray. The book mentions possible business strategies of such publishers as Encyclopædia Britannica who have to confront how to stay viable as technology changes the value and availability of information.
BERT (language model)
Bidirectional encoder representations from transformers (BERT) is a language model introduced in October 2018 by researchers at Google. It learns to represent text as a sequence of vectors using self-supervised learning. It uses the encoder-only transformer architecture. BERT dramatically improved the state of the art for large language models. As of 2020, BERT is a ubiquitous baseline in natural language processing (NLP) experiments. BERT is trained by masked token prediction and next sentence prediction. With this training, BERT learns contextual, latent representations of tokens in their context, similar to ELMo and GPT-2. It found applications for many natural language processing tasks, such as coreference resolution and polysemy resolution. It improved on ELMo and spawned the study of "BERTology", which attempts to interpret what is learned by BERT. BERT was originally implemented in the English language at two model sizes, BERTBASE (110 million parameters) and BERTLARGE (340 million parameters). Both were trained on the Toronto BookCorpus (800M words) and English Wikipedia (2,500M words). The weights were released on GitHub. On March 11, 2020, 24 smaller models were released, the smallest being BERTTINY with just 4 million parameters. == Architecture == BERT is an "encoder-only" transformer architecture. At a high level, BERT consists of 4 modules: Tokenizer: This module converts a piece of English text into a sequence of integers ("tokens"). Embedding: This module converts the sequence of tokens into an array of real-valued vectors representing the tokens. It represents the conversion of discrete token types into a lower-dimensional Euclidean space. Encoder: a stack of Transformer blocks with self-attention, but without causal masking. Task head: This module converts the final representation vectors into one-shot encoded tokens again by producing a predicted probability distribution over the token types. It can be viewed as a simple decoder, decoding the latent representation into token types, or as an "un-embedding layer". The task head is necessary for pre-training, but it is often unnecessary for so-called "downstream tasks," such as question answering or sentiment classification. Instead, one removes the task head and replaces it with a newly initialized module suited for the task, and finetune the new module. The latent vector representation of the model is directly fed into this new module, allowing for sample-efficient transfer learning. === Embedding === This section describes the embedding used by BERTBASE. The other one, BERTLARGE, is similar, just larger. The tokenizer of BERT is WordPiece, which is a sub-word strategy like byte-pair encoding. Its vocabulary size is 30,000, and any token not appearing in its vocabulary is replaced by [UNK] ("unknown"). The first layer is the embedding layer, which contains three components: token type embeddings, position embeddings, and segment type embeddings. Token type: The token type is a standard embedding layer, translating a one-hot vector into a dense vector based on its token type. Position: The position embeddings are based on a token's position in the sequence. BERT uses absolute position embeddings, where each position in a sequence is mapped to a real-valued vector. Each dimension of the vector consists of a sinusoidal function that takes the position in the sequence as input. Segment type: Using a vocabulary of just 0 or 1, this embedding layer produces a dense vector based on whether the token belongs to the first or second text segment in that input. In other words, type-1 tokens are all tokens that appear after the [SEP] special token. All prior tokens are type-0. The three embedding vectors are added together representing the initial token representation as a function of these three pieces of information. After embedding, the vector representation is normalized using a LayerNorm operation, outputting a 768-dimensional vector for each input token. After this, the representation vectors are passed forward through 12 Transformer encoder blocks, and are decoded back to 30,000-dimensional vocabulary space using a basic affine transformation layer. === Architectural family === The encoder stack of BERT has 2 free parameters: L {\displaystyle L} , the number of layers, and H {\displaystyle H} , the hidden size. There are always H / 64 {\displaystyle H/64} self-attention heads, and the feed-forward/filter size is always 4 H {\displaystyle 4H} . By varying these two numbers, one obtains an entire family of BERT models. For BERT: the feed-forward size and filter size are synonymous. Both of them denote the number of dimensions in the middle layer of the feed-forward network. the hidden size and embedding size are synonymous. Both of them denote the number of real numbers used to represent a token. The notation for encoder stack is written as L/H. For example, BERTBASE is written as 12L/768H, BERTLARGE as 24L/1024H, and BERTTINY as 2L/128H. == Training == === Pre-training === BERT was pre-trained simultaneously on two tasks: Masked language modeling (MLM): In this task, BERT ingests a sequence of words, where one word may be randomly changed ("masked"), and BERT tries to predict the original words that had been changed. For example, in the sentence "The cat sat on the [MASK]," BERT would need to predict "mat." This helps BERT learn bidirectional context, meaning it understands the relationships between words not just from left to right or right to left but from both directions at the same time. Next sentence prediction (NSP): In this task, BERT is trained to predict whether one sentence logically follows another. For example, given two sentences, "The cat sat on the mat" and "It was a sunny day", BERT has to decide if the second sentence is a valid continuation of the first one. This helps BERT understand relationships between sentences, which is important for tasks like question answering or document classification. ==== Masked language modeling ==== In masked language modeling, 15% of tokens would be randomly selected for masked-prediction task, and the training objective was to predict the masked token given its context. In more detail, the selected token is: replaced with a [MASK] token with probability 80%, replaced with a random word token with probability 10%, not replaced with probability 10%. The reason not all selected tokens are masked is to avoid the dataset shift problem. The dataset shift problem arises when the distribution of inputs seen during training differs significantly from the distribution encountered during inference. A trained BERT model might be applied to word representation (like Word2Vec), where it would be run over sentences not containing any [MASK] tokens. It is later found that more diverse training objectives are generally better. As an illustrative example, consider the sentence "my dog is cute". It would first be divided into tokens like "my1 dog2 is3 cute4". Then a random token in the sentence would be picked. Let it be the 4th one "cute4". Next, there would be three possibilities: with probability 80%, the chosen token is masked, resulting in "my1 dog2 is3 [MASK]4"; with probability 10%, the chosen token is replaced by a uniformly sampled random token, such as "happy", resulting in "my1 dog2 is3 happy4"; with probability 10%, nothing is done, resulting in "my1 dog2 is3 cute4". After processing the input text, the model's 4th output vector is passed to its decoder layer, which outputs a probability distribution over its 30,000-dimensional vocabulary space. ==== Next sentence prediction ==== Given two sentences, the model predicts if they appear sequentially in the training corpus, outputting either [IsNext] or [NotNext]. During training, the algorithm sometimes samples two sentences from a single continuous span in the training corpus, while at other times, it samples two sentences from two discontinuous spans. The first sentence starts with a special token, [CLS] (for "classify"). The two sentences are separated by another special token, [SEP] (for "separate"). After processing the two sentences, the final vector for the [CLS] token is passed to a linear layer for binary classification into [IsNext] and [NotNext]. For example: Given "[CLS] my dog is cute [SEP] he likes playing [SEP]", the model should predict [IsNext]. Given "[CLS] my dog is cute [SEP] how do magnets work [SEP]", the model should predict [NotNext]. === Fine-tuning === BERT is meant as a general pretrained model for various applications in natural language processing. That is, after pre-training, BERT can be fine-tuned with fewer resources on smaller datasets to optimize its performance on specific tasks such as natural language inference and text classification, and sequence-to-sequence-based language generation tasks such as question answering and conversational response generation. The original BERT paper published results demonstrating that a small amount of fine
Elonis v. United States
Elonis v. United States, 575 U.S. 723 (2015), was a United States Supreme Court case concerning whether conviction of threatening another person over interstate lines (under 18 U.S.C. § 875(c)) requires proof of subjective intent to threaten or whether it is enough to show that a "reasonable person" would regard the statement as threatening. In controversy were the purported threats of violent rap lyrics written by Anthony Douglas Elonis and posted to Facebook under a pseudonym. The ACLU filed an amicus brief in support of the petitioner. It was the first time the Court has heard a case considering true threats and the limits of speech on social media. == Background == In May 2010, Elonis was in the process of divorce and made a number of public Facebook posts. Prior to his postings, he had lost his job at an amusement park. He "posted the script of a sketch" by The Whitest Kids U' Know, which originally referenced saying "I want to kill the President of the United States" and replaced the president with his wife: Elonis ended the post with this statement: "Art is about pushing limits. I'm willing to go to jail for my constitutional rights. Are you?" A week later, Elonis posted about local law enforcement and a kindergarten class, which caught the attention of the Federal Bureau of Investigation. Then, he wrote a post on Facebook about one of the agents who visited him: He concluded: == Arrest and Conviction == These actions led to Elonis's arrest on December 8, 2010. He was indicted by a grand jury on five counts of threats to his estranged ex-wife, park employees and visitors, local law enforcement, an FBI agent, and a kindergarten class that had been relayed through interstate communication. At the district court, Elonis moved to dismiss the indictment for failing to allege that he had intended to threaten anyone, claiming his Facebook post was not were not intended as a threat. He argued that, as an aspiring rap artist, his posts were intended to be a form of artistic expression to help him cope with his recent loses. According to him, he did not mean anything said in his posts in a literal sense. His motion was denied. He requested a jury instruction that "the government must prove that he intended to communicate a true threat", which was also denied. He was convicted on the last four of the five counts, and was sentenced to 44 months in prison and three years on supervised release. He appealed unsuccessfully to the Third Circuit, renewing his challenge to the jury instructions. He then appealed to the U.S. Supreme Court based on lack of any attempt to show intent to threaten and on First Amendment rights. == Decision == On June 1, 2015, the U.S. Supreme Court reversed Elonis's conviction in an 8–1 decision. Chief Justice John Roberts wrote for a seven-justice majority, Samuel Alito authored an opinion concurring in part and dissenting in part, and Clarence Thomas authored a dissenting opinion. The finding of the circuit court was reversed and the matter remanded. === Majority opinion === The majority opinion, written by Roberts, did not rule on First Amendment matters or on the question of whether recklessness was sufficient mens rea to show intent. It ruled that mens rea was required to prove the commission of a crime under §875(c). Importantly, the mens rea issue had been preserved for review, since Elonis had raised that objection at every stage of the previous proceedings. The government contended that the presence of the words "intent to extort" in §875(b) and §875(d) implied that the absence in §875(c) was constructive. The court disagreed, holding that the absence of the language in §875(c) was because the section was intended to have a broader scope than threats relating to extortion. The opinion drew on many Supreme Court cases holding that in criminal law, mens rea was required though it had not been mentioned explicitly in statute. Consequently, the Supreme Court ruled in favor of Elonis. === Alito's concurrence === Justice Samuel Alito, concurring in part and dissenting in part, opined that while agreeing that mens rea was required and specifically that showing negligence was not sufficient, the court should have ruled on the question of recklessness. He further opined that recklessness was sufficient to show a crime under that provision on the basis that going further would amount to amending the statute, rather than interpreting it. Since Elonis explicitly argued that recklessness was not sufficient, Alito said: I would therefore remand for the Third Circuit to determine if Elonis’s failure (indeed, refusal) to argue for recklessness prevents reversal of his conviction. The Third Circuit should also have the opportunity to consider whether the conviction could be upheld on harmless error grounds. Alito also addressed the First Amendment question, elided by the majority opinion. He held that "lyrics in songs that are performed for an audience or sold in recorded form are unlikely to be interpreted as a real threat to a real person. ... Statements on social media that are pointedly directed at their victims, by contrast, are much more likely to be taken seriously." === Thomas's dissent === Justice Clarence Thomas, dissenting, wrote against discarding the "general intent" standard without replacing it with a clearer standard. Thomas argued that "there is no historical practice requiring more than general intent when a statute regulates speech." Thomas cited Rosen v. United States, arguing that general intent was sufficient in this case. However, the majority opinion offers refutation in that Rosen turned on ignorance of the law: knowledge as to whether material was legally obscene, not on whether it was intended to be obscene. Thomas also supported the government's claim that the presence of "intent to extort" language in the adjacent §875(b) and did not address the majority's reasoning on that language. Thomas used precedent, notably from the states and 18th-century England based on other but similar and, arguably, influencing legislation to support his "general intent" claim. Thomas also drew a parallel with general intent in tort. While he sought to address the First Amendment issues, he never strayed far from "general intent". == Aftermath == On remand, the Third Circuit reaffirmed the conviction "concluding beyond a reasonable doubt that Elonis would have been convicted if the jury had been properly instructed" and therefore was harmless error. In 2022, Elonis was once again arrested and indicted on three counts of cyberstalking involving three people. It was discovered that between 2018 and 2021, Elonis had sent numerous threatening messages over email, text, voice mail, and social media platforms like Twitter to a former prosecutor of the Eastern District of Pennsylvania, his ex-girlfriend, and ex-wife. On August 5, after a five-day trial, Elonis was found guilty on all three counts, and on March 23, 2023, he was sentenced by U.S. District Court Judge Edward G. Smith of Easton, Pennsylvania to twelve years and seven months in prison.
Quality of experience
Quality of experience (QoE) is a measure of the delight or annoyance of a customer's experiences with a service (e.g., web browsing, phone call, TV broadcast). QoE focuses on the entire service experience; it is a holistic concept, similar to the field of user experience, but with its roots in telecommunication. QoE is an emerging multidisciplinary field based on social psychology, cognitive science, economics, and engineering science, focused on understanding overall human quality requirements. == Definition and concepts == In 2013, within the context of the COST Action QUALINET, QoE has been defined as:The degree of delight or annoyance of the user of an application or service. It results from the fulfillment of his or her expectations with respect to the utility and / or enjoyment of the application or service in the light of the user’s personality and current state.This definition has been adopted in 2016 by the International Telecommunication Union in Recommendation ITU-T P.10/G.100. Before, various definitions of QoE had existed in the domain, with the above-mentioned definition now finding wide acceptance in the community. QoE has historically emerged from Quality of Service (QoS), which attempts to objectively measure service parameters (such as packet loss rates or average throughput). QoS measurement is most of the time not related to a customer, but to the media or network itself. QoE however is a purely subjective measure from the user's perspective of the overall quality of the service provided, by capturing people's aesthetic and hedonic needs. QoE looks at a vendor's or purveyor's offering from the standpoint of the customer or end user, and asks, "What mix of goods, services, and support, do you think will provide you with the perception that the total product is providing you with the experience you desired and/or expected?" It then asks, "Is this what the vendor/purveyor has actually provided?" If not, "What changes need to be made to enhance your total experience?" In short, QoE provides an assessment of human expectations, feelings, perceptions, cognition and satisfaction with respect to a particular product, service or application. QoE is a blueprint of all human subjective and objective quality needs and experiences arising from the interaction of a person with technology and with business entities in a particular context. Although QoE is perceived as subjective, it is an important measure that counts for customers of a service. Being able to measure it in a controlled manner helps operators understand what may be wrong with their services and how to improve them. == QoE factors == QoE aims at taking into consideration every factor that contributes to a user's perceived quality of a system or service. This includes system, human and contextual factors. The following so-called "influence factors" have been identified and classified by Reiter et al.: Human Influence Factors Low-level processing (visual and auditory acuity, gender, age, mood, …) Higher-level processing (cognitive processes, socio-cultural and economic background, expectations, needs and goals, other personality traits…) System Influence Factors Content-related Media-related (encoding, resolution, sample rate, …) Network-related (bandwidth, delay, jitter, …) Device-related (screen resolution, display size, …) Context Influence Factors Physical context (location and space) Temporal context (time of day, frequency of use, …) Social context (inter-personal relations during experience) Economic context Task context (multitasking, interruptions, task type) Technical and information context (relationship between systems) Studies in the field of QoE have typically focused on system factors, primarily due to its origin in the QoS and network engineering domains. Through the use of dedicated test laboratories, the context is often sought to be kept constant. == QoE versus User Experience == QoE is strongly related to but different from the field of User Experience (UX), which also focuses on users' experiences with services. Historically, QoE has emerged from telecommunication research, while UX has its roots in Human–Computer Interaction. Both fields can be considered multi-disciplinary. In contrast to UX, the goal of improving QoE for users was more strongly motivated by economic needs. Wechsung and De Moor identify the following key differences between the fields: == QoE measurement == As a measure of the end-to-end performance at the service level from the user's perspective, QoE is an important metric for the design of systems and engineering processes. This is particularly relevant for video services because – due to their high traffic demands –, bad network performance may highly affect the user's experience. So, when designing systems, the expected output, i.e. the expected QoE, is often taken into account – also as a system output metric and optimization goal. To measure this level of QoE, human ratings can be used. The mean opinion score (MOS) is a widely used measure for assessing the quality of media signals. It is a limited form of QoE measurement, relating to a specific media type, in a controlled environment and without explicitly taking into account user expectations. The MOS as an indicator of experienced quality has been used for audio and speech communication, as well as for the assessment of quality of Internet video, television and other multimedia signals, and web browsing. Due to inherent limitations in measuring QoE in a single scalar value, the usefulness of the MOS is often debated. Subjective quality evaluation requires a lot of human resources, establishing it as a time-consuming process. Objective evaluation methods can provide quality results faster, but require dedicated computing resources. Since such instrumental video quality algorithms are often developed based on a limited set of subjective data, their QoE prediction accuracy may be low when compared to human ratings. QoE metrics are often measured at the end devices and can conceptually be seen as the remaining quality after the distortion introduced during the preparation of the content and the delivery through the network, until it reaches the decoder at the end device. There are several elements in the media preparation and delivery chain, and some of them may introduce distortion. This causes degradation of the content, and several elements in this chain can be considered as "QoE-relevant" for the offered services. The causes of degradation are applicable for any multimedia service, that is, not exclusive to video or speech. Typical degradations occur at the encoding system (compression degradation), transport network, access network (e.g., packet loss or packet delay), home network (e.g. WiFi performance) and end device (e.g. decoding performance). == QoE management == Several QoE-centric network management and bandwidth management solutions have been proposed, which aim to improve the QoE delivered to the end-users. When managing a network, QoE fairness may be taken into account in order to keep the users sufficiently satisfied (i.e., high QoE) in a fair manner. From a QoE perspective, network resources and multimedia services should be managed in order to guarantee specific QoE levels instead of classical QoS parameters, which are unable to reflect the actual delivered QoE. A pure QoE-centric management is challenged by the nature of the Internet itself, as the Internet protocols and architecture were not originally designed to support today's complex and high demanding multimedia services. As an example for an implementation of QoE management, network nodes can become QoE-aware by estimating the status of the multimedia service as perceived by the end-users. This information can then be used to improve the delivery of the multimedia service over the network and proactively improve the users' QoE. This can be achieved, for example, via traffic shaping. QoE management gives the service provider and network operator the capability to minimize storage and network resources by allocating only the resources that are sufficient to maintain a specific level of user satisfaction. As it may involve limiting resources for some users or services in order to increase the overall network performance and QoE, the practice of QoE management requires that net neutrality regulations are considered.
Vinyl cutter
A vinyl cutter is an entry-level machine for making signs. Computer-designed vector files with patterns and letters are directly cut on the roll of vinyl which is mounted and fed into the vinyl cutter through USB or serial cable. Vinyl cutters are mainly used to make signs, banners and advertisements. Advertisements seen on automobiles and vans are often made with vinyl cut letters. While these machines were designed for cutting vinyl, they can also cut through computer and specialty papers, as well as thicker items like thin sheets of magnet. In addition to sign business, vinyl cutters are commonly used for apparel decoration. To decorate apparel, a vector design needs to be cut in mirror image, weeded, and then heat applied using a commercial heat press or a hand iron for home use. Some businesses use their vinyl cutter to produce both signs and custom apparel. Many crafters also have vinyl cutters for home use. These require little maintenance, and the vinyl can be bought in bulk relatively cheaply. Vinyl cutters are also often used by stencil artists to create single use or reusable stencil art and lettering == How it works == A vinyl cutter is a type of computer-controlled machine tool. The computer controls the movement of a sharp blade over the surface of the material as it would the nozzles of an ink-jet printer. This blade is used to cut out shapes and letters from sheets of thin self-adhesive plastic (vinyl). The vinyl can then be stuck to a variety of surfaces depending on the adhesive and type of material. To cut out a design, a vector-based image must be created using vector drawing software. Some vinyl cutters are marketed to small in-home businesses and require download and use of a proprietary editing software. The design is then sent to the cutter where it cuts along the vector paths laid out in the design. The cutter is capable of moving the blade on an X and Y axis over the material, cutting it into the required shapes. The vinyl material comes in long rolls allowing projects with significant length like banners or billboards to be easily cut. A major limitation with vinyl cutters is that they can only cut shapes from solid colours of vinyl, paper, card or thin plastic sheets such as Mylar. The type and thickness of material will vary for each cutter and how much downforce the cutter is capable of. If the material has no backing, a backing sheet, material or cutting mat and a temporary adhesive are needed to allow the cutter to cut through the material. A design with multiple colours must have each colour cut separately and then layered on top of each other as it is applied to the substrate. This is a process that is often applied in stencil art. Also, since the shapes are cut out of solid colours, photographs and gradients cannot be reproduced with a stand-alone cutter. === Design creation === Designs are created using vector-based software like Adobe Illustrator, FlexiSign, EasyCutPro, or other software. Vector artwork is either drawn with lines, shapes and text or images are vectorized thus create vector shapes. Most cutters (also called plotters) require special software to load/edit the artwork and communicate with the cutter. Computer designed images are loaded onto the vinyl cutter via a wired connection or over a wireless protocol. Then the vinyl is loaded into the machine where it is automatically fed through and cut to follow the set design. The vinyl can be placed on an adhesive mat to stabilize the vinyl when cutting smaller designs. === Types of vinyl === Adhesive vinyl is the type of vinyl used for store windows, car decals, signage, and more. Adhesive vinyl is applied with a transfer medium often called "transfer tape" or "carrier sheet". Heat transfer vinyl is the type of vinyl used to apply a design to fabric including t-shirts, tea towels, canvas bags, and more. Heat Transfer vinyl can be applied using a heat press or an iron, though the constant pressure and heat from a heat press is recommended by experts. === Using other materials === In addition to vinyl some cutters are capable of cutting other materials such as paper, card, plastic sheets and even thin wood. The thickness and type of material that can be cut will depend on the model of the cutter and heavily depends on the downforce. Cricut is a popular home cutter used by arts and craft enthusiasts since it allows for a wide use of different materials and is similar in size to a household printer and has strong downforce for its size. === Backing and cutting mat === If you cut material that doesn't have an adhesive backing you will require a cutting mat that you need to attach your material to. Some cutting mats are sticky, others will require you to use a temporary adhesive and/or masking tape to keep the material in place when cutting. === Cutting === The vinyl cutter uses a small knife or blade to precisely cut the outline of figures into a sheet or piece of vinyl, but not the release liner. The process of cutting vinyl material without penetrating it completely is referred to as "kiss cutting". The knife moves side to side and turns, while the vinyl is moved beneath the knife. The results from the cut process is an image cut into the material. === Weeding === The material is then 'weeded' where the excess parts of the figures are removed from the release liner. It is possible to remove the positive parts, which would give a negative decal, or remove the negative parts, giving a positive decal. Removing the figure would be like removing the positive, giving a negative image of the figures. === Transfer tape === A sheet of transfer tape with an adhesive backing is laid on the weeded vinyl when necessary. Heat Transfer vinyl often does not require use of a separate transfer tape. A roller is applied to the tape, causing it to adhere to the vinyl. The transfer tape and the weeded vinyl is pulled off the release liner, and applied to a substrate, such as a sheet of aluminium. This results in an aluminium sign with vinyl figures. == Uses == In addition to the capabilities of the cutter itself, adhesive vinyl comes in a wide variety of colors and materials including gold and silver foil, vinyl that simulates frosted glass, holographic vinyl, reflective vinyl, thermal transfer material, and even clear vinyl embedded with gold leaf. (Often used in the lettering on fire trucks and rescue vehicles.) As the vinyl film is supplied by the manufacturer, it comes attached to a release liner. == Challenges when cutting on a vinyl cutter == Cutting on a vinyl cutter requires careful calibration to achieve clean and accurate results, especially when the goal is to cut through only the top layer of material while leaving the backing intact. One of the most common challenges is setting the correct cutting depth. If the blade is not lowered enough, the vinyl material may not separate properly; if it goes too deep, it can cut through the backing layer and potentially damage the cutting mat. The cutting depth on the vinyl cutter machines typically does not exceed 1 mm. Another frequent issue is the mismatch between the blade and the type of material being processed. Using an inappropriate blade can lead to uneven cuts, premature dulling of the edge, and torn or frayed material. The overall quality of the output also depends on factors such as the cutting speed, blade sharpening and cutting angle, and the material the knife is made of.
Circular thresholding
Circular thresholding is an algorithm for automatic image threshold selection in image processing. Most threshold selection algorithms assume that the values (e.g. intensities) lie on a linear scale. However, some quantities such as hue and orientation are a circular quantity, and therefore require circular thresholding algorithms. The example shows that the standard linear version of Otsu's method when applied to the hue channel of an image of blood cells fails to correctly segment the large white blood cells (leukocytes). In contrast the white blood cells are correctly segmented by the circular version of Otsu's method. == Methods == There are a relatively small number of circular image threshold selection algorithms. The following examples are all based on Otsu's method for linear histograms: (Tseng, Li and Tung 1995) smooth the circular histogram, and apply Otsu's method. The histogram is cyclically rotated so that the selected threshold is shifted to zero. Otsu's method and histogram rotation are applied iteratively until several heuristics involving class size, threshold location, and class variance are satisfied. (Wu et al. 2006) smooth the circular histogram until it contains only two peaks. The histogram is cyclically rotated so that the midpoint between the peaks is shifted to zero. Otsu's method and histogram rotation are applied iteratively until convergence of the threshold. (Lai and Rosin 2014) applied Otsu's method to the circular histogram. For the two class circular thresholding task they showed that, for a histogram with an even number of bins, the optimal solution for Otsu's criterion of within-class variance is obtained when the histogram is split into two halves. Therefore the optimal solution can be efficiently obtained in linear rather than quadratic time. == References and further reading == D.-C. Tseng, Y.-F. Li, and C.-T. Tung, Circular histogram thresholding for color image segmentation in Proc. Int. Conf. Document Anal. Recognit., 1995, pp. 673–676. J. Wu, P. Zeng, Y. Zhou, and C. Olivier, A novel color image segmentation method and its application to white blood cell image analysis in Proc. Int. Conf. Signal Process., vol. 2. 2006, pp. 16–20. Y.K. Lai, P.L. Rosin, Efficient Circular Thresholding, IEEE Trans. on Image Processing 23(3), 992–1001 (2014). doi:10.1109/TIP.2013.2297014
Institute of Telecommunications Professionals
The Institute of Telecommunications Professionals (ITP) is a membership organisation for professionals in the telecommunications industry, based in the United Kingdom. The Institute was originally founded in 1906. It is now a registered company with Companies House in the United Kingdom, incorporated in 2002. Brendan O' Mahony has been the chief executive of the ITP. Lucy Woods presided over ITP for fifteen years, until 2018, when the organization named Kevin Paige chairman for five years. In 2022 the ITP appointed its new CEO, Charlotte Goodwill. In 2021, the ITP assisted a UK fibre network Vorboss in establishing its training academy. In 2023, the ITP appointed Tim Creswick, the CEO of Vorboss, as the new chair of its board of directors. The institute has an associated journal, the Journal of the Institute of Telecommunications Professionals, established in 2007 and published quarterly.