MATO is a USB drive, and it is a fantastic USB drive, not only offers a feature that simple, compared to the shape of the outer ones are almost u MATO don’t know it can throw in a few streets. In a tense spare time you can use MATO to casually around a variety of shapes to delight in this, or is conveniently winding it back in your bag or clothes as an accessory to match today’s mood. Vans iPhone 6 Case
MATO fabric is similar to the skin and give you a smooth feeling, makes you feel like to touch it. MATO is a bit like a snake, wrapped around your body. Imagine, you need to use a USB memory stick when you hand in the eyes of others from your hand ring down plug in computer to copy the data back, family moments, put will you u so handsome.
There are many colors to choose from and MATO has, MATO uses the latest COB (chip on board) USB Flash memory chips. And MATO is waterproof, dust-proof and anti-static, so you don’t have to worry about accidentally used to MATO was damaged. Vans iPhone
You can with a plastic bracelet with MATO to run or rainy outdoor activities OK. It has 6 possible combinations of color, has a memory capacity of 8GB and 16GB to choose from, like it?
MAOTO consists of 4 main parts: the outer layers of the fabric, springs, Flash drives and casing. Textile and rigid helmet providing in Denmark, springs in Italy, Flash drive bulk order from Silicon Valley companies, shell injection molding in China. All components will be shipped to Italy in Milan, where they will be assembled, packaged, shipped last. If you like you can go to Kickstarter for their sponsorship.
Apple keyboard years unchanged, want to play differently, maybe you can try this on Kickstarter wood keyboard Macbook. This keyboard is not a full keyboard, a keyboard made of wood it is posted, you can post it on the Mac computer keyboards, so that we can give you a unique and very tactile keyboard.
This keyboard is located in Brooklyn, New York, a design studio called RAW. Keyboard has two kinds of material can be selected, one is rosewood, other is bamboo, is a relatively common wood. Disney Note 4 flip case
RAW Studio, rosewood may provide a deep and rich, smooth texture. And the bamboo gives a feeling of pure.
RAW Studio, they think the wooden keyboard most of the quality and handling of the product details. Unlike adhesive keyboards will leave traces after the tear, so they go through extensive testing, and finally decided to use 3M products.
This keyboard manufacturing process need to Polish the wood, making the wood more smooth.
Of course you don’t have to worry about with this keyboard will make big changes in the thickness of the keyboard, look at the pictures, they are very thin.
Next process is that the keyboard is laser cut, finish to polished with sandpaper after cutting. Of course the most important thing is to paint on the keyboard.
RAW will be processed for each button hole, it can perfectly match the Macbook keyboard backlight.
In addition to wooden keyboard Macbook, RAW Studio regular Mac keyboard versions. Disney Galaxy Note 4 case
This keyboard sells for $40. Favorite friends can click here to make a purchase.
The simplest design is the most fashion. Designer Shel Han recently launched a group called the “knock-down” DIY furniture, minimalist garden wind and the ease of combination will immediately catch your eye. Shel Han inspiration from early China good joiner, architect, knock-down design without using any screws and glue can easily assemble furniture, sounds exciting. Jeremy Scott iPad Mini Case
Knock-down used painted wood for a unique material, contains four main types of components, connection using the original holes fitted with mustard instead of screws and glue, so you don’t need to use too much, you can DIY combinations 10 kinds of furniture out of it. Small armchair, stool, table, shelves … … When inserting a, let your self-confidence bursting to the extreme! This pride may date back to the age of four building blocks of time.
We are a team of VR design, mainly responsible for design of VR products. This is our team doing things, explore and build a more complete VR design process, focusing on design tools for research and job analysis, optimization of all processes in order to improve team productivity and output quality of VR. In actual VR projects continues to be optimized, but also welcome all ideas and suggestions.
Second, what to do?
Sort by priority, is divided into four parts:
1, work processes, each role needs to be concerned and the Division of labour within the team.
2, the use of design tools.
3, user research methods, demand-side management.
4 design principles (design) induction and establishment.
As shown in the following figure:
Following is a detailed description of the content of four parts.
| Clear division of duties
1, establish workflow: VR design all the key aspects of the process and work together for the combing, draw a flowchart.
2, combs work: mainly include process various part of staff responsibilities, the main outputs, with.
Explicit VR project design in the various functional areas of responsibility, of output. All output through practical projects and gradually establish a standard template, project process standardization.
| Using design tools
Study 1, and what tools to use to design and try out.
2, the main tool for staff training, mastering the basic use of the tool.
Why all the designers have to learn 3D design tools?
Traditional dimensional design software (such as Axure) is not quick and easy display of 3D spaces product design, in the limit of 2D do 3D stuff, complicated process and there is no way iteration.
Design Master-d main tools foundations use is necessary.
How to set up spatial thinking?
The best way is to use 3D design software to design, traditional 2D designer during the design process can be a good idea to broaden the space for performances. In addition, the game engine (Unity) of basic learning, using, can also help you learn to understand better 3D gaming, VR product design and implementation principle to avoid designer’s content could not be achieved.
| Create demand for research and pool
Research needs pool set: VR user studies and VR user needs and establishing the framework and content planning, combing IJB processes and requirements pool template.
Why demand for pools?
Requirement capture and transformation of product design is an important work, in the course of our user research, often refined and thinking of weakening demand, think the Institute needs is real user feedback in user demand. But the demand for research and design are common, in fact, is the blending of water and soil.
Create demand for pool, virtual reality can be more targeted information what users want, by user portraits, stories, research tools, such as access to the pool needs into demand, needs finishing, precipitation, enable us to quickly and accurately design the product in the late.
| Design principles, norms
Design of normative framework has been developed: establish VR design framework, enumerated VR design specification content and direction.
Current VR product much less sophisticated experience. We know the product experience, the most important point is to maintain standardization and uniformity. Virtual reality products is different from traditional Internet products, designers concerned about more than just Visual impact on the users, in addition, sound, touch, space control VR the user experience will have a big impact.
Available interactive specification specification for interactive design of Cardboard with Google only VR for mobile end devices.
Therefore we hope that VR user experience studies and research, summary and to find out what the principles are good, fit VR class of products.
Preliminary thoughts on the direction of specification mainly involve:
Follow-up task is, to establish norms in all directions and templates, to which gradually fills, continue iteration.
Establishment of the standard component library, and reuse the content of the same design.
Third, the clear assignment of responsibility
Pick several major link below explaining the content of the work, in a later post I will start to every aspect of content detail.
1, functional design. What products should be achieved, functional business what is the logic behind.
2, scenario planning. List divided into different scenes, the output list.
Building 3, VR scenes. Required functionality and business logic for each scenario describe, draw 2D plans, figure to be included in all objects in the current scene.
According to the design and the fine arts 3D resources, output to the development process.
Final development for full product functionality. Space is limited, in future articles, the next part of my paper did describe in detail.
Lei feng’s network (search for “Lei feng’s network”, public interest) by the writer shanghongwei, net dragons Web VR interaction designer of the company.
You’ve ever heard before King Jim? eh, this sounds a lot like cigarette producer company is Japan Office supplies company, specially introduced a wide variety of interesting equipment. For example the earlier King Jim launched the integrated mouse and camera equipment. Well, very interesting. Valentino iPhone 6s cases
This time, King Jim brings an ear, with vibration function “alarm clock”-Mezamashi (alarm) Earphone NMR10. This NMR10 can be shaken to prompt the user, able to work under the premise of not disturbing others woke up from his sleep.
In General, there are many commuters in the morning commuter car or desk take a NAP in the afternoon. Usually these moments no alarm bells probably can’t wake, but set the alarm will influence others. NMR10 hands at this point.
NMR10 is essentially a standard headset, compatible with a variety of smart phones, and other music devices. Meanwhile, NMR10 has built-in vibrating parts, used as an alarm to alert users to get up. Users can choose according to their actual needs strong, medium, weak three different vibration strengths. The headphones do not need any external power supply-driven, and need only one AAA battery for power supply control panel. Full State AAA battery life of up to six months at a time. Valentino i6 plus case
It is reported that the final selling price of this NMR10 phone will reach us $ 53 (about 343 Yuan), you like small partners may wish to take a look at it.
Foreign media said the following after Google, Baidu, Russia Yandex search engine giant also announced recently that development of self-driving cars. Victorias Secret Case
Yandex and the Russian truck manufacturer Kamaz jointly develop small and medium sized automatic driving buses, the automatic driving small electric bus can carry up to 12 passengers, capable of traveling 200 kilometers per charge. Yandex resources into the field of computer vision, speech recognition, algorithm, while the Kamaz automobile manufacturing resources provides entities. Kamaz has strong background, by the Russian Government, Germany’s Daimler invested, as well as set up by the Russian Government to support NAMI automobile Research Institute.
As the search engine giants, Yandex it network services integrated into the in-vehicle systems. Earlier Yandex and companies such as Toyota and Honda have maintained cooperation, integration of car sales in Russia has its own navigation and Web browsing services.
Self-driving cars have become the craze, traditional companies and new technology companies rush to invest substantial human and financial resources for research and development, most enterprises plans to launch products at around 2020. Yandex entered relatively late, there was no give much time to market. According to the NAMI Research Institute said the road test plans at the beginning of 2017.
Google, Baidu no car is in beta, but has accumulated a lot of skill and experience. Meanwhile, Yandex money are not strong companies such as Google, Baidu, second-quarter earnings Yandex income of US $ 280 million, net income of $ 32 million. Yandex in the technical, financial, timing the premise of three important components were at a disadvantage, in order to catch up with the former also needs a certain amount of time.
Coming in September (traditionally should be) released iPhone 7, is one of the largest and most long rumored before every iPhone are the standard 3.5mm Headphone Jack will be cancelled, replaced by Lightning interface as a headphone port, users can also use a Bluetooth wireless headset or wired headphones connector continues to use the 3.5mm interface.
Lei Feng network (search “Lei Feng network” public, concern) small series and no enough of message source to confirmed this a rumors, but industry almost are has finds Apple will will so do, which except for Rob in Apple zhiqian became first canceled headphones interface manufacturers of Le depending on zhiwai, also has August 20 for media held has a small new tasting will of crazybaby (mad child technology). iPhone6 Vans
Crazybaby familiar they had launched a magnetic speaker Mars, appeared before the new iPhone is released-the–Air is a wireless headset by crazybaby. And like an old subsection, new products have a lot to talk about the unique design and tonality.
First for wireless headphones carry not convenient, put in package in hard to find of problem, crazybaby for Air design has a portable capsule, this shape “dirt dirt da” anode alumina capsule than 4.7 inches of iPhone long some, internal is to placed headphones of magnetic sucking type charging base; external has a to capsule charging of USB Type-C interface, and twist two or three Xia on can open of cover.
Capsule with a battery you can charge the headset provides you with an additional 4-opportunities, just 1 hour per charge, and makes the headset’s battery life up to 16 hours. Small series if accession to the vibration function should be sold (Atlantic Ocean). iPhone6 Vans
Headphone connection, Air is a separate pair of wireless headphones, 4.2 connecting via Bluetooth mobile phones. Bluetooth 4.2 compared with previous-generation data transmission up to 2.5 times times, 10 times the data capacity, Air also introduced the IPV6 Protocol, reduced power consumption and safety.
A button on the headset, you can implement features include play/pause music, telephone answering, awake Siri, features such as volume control, because there is only one button operation realized by the press as well as by official eventually accompanied by a guide.
As products are still used to experience engineering machines, unable to work, wearing comfort and sound quality, the next conclusion. Crazybaby said the final version will be a small circle, smaller ears can adapt to the Asian girl.
Sound quality, Air file transfer via data interface to headphones, earphones to take over decoding, follow-up work at the corresponding. Its equipped with HD resolution CSR chip, while the use of True TWS Stereo stereo.
Air Mobile and we have a crazybaby app, you can control the Air and make further settings such as EQ.
Crazybaby on this and did not release the price of Air, only that standards relative to the product, the price will be lower. Listed until this Christmas.
In small’s view, if only to introduce a wireless headset, obviously cannot be said to be betting on Apple canceling headphone jack, after all, this is a mature category, there is no 3.5mm headphone jack can be used.
Crazybaby domestic audio as a new brand, and Wang Feng FILL headset face find its place in the market. FILL had previously said its product sales of 75,000 units, crazybaby released a product sales to 50,000 units, most of which come from areas outside the city.
Crazybaby founder Allen said about it, “as long as the product works, foreign users are usually more receptive to new brands, while domestic users may consider some more.” Headphones are such a great value brand in street markets, crazybaby did not intend to hit big money stars like the Beats, but hope that more users will be propagated.
In the escalation of domestic consumption trend, more and more domestic independent brands supported by the emerging middle class, crazybaby is probably the one.
Lei feng’s network (search for “Lei feng’s network” public attention) by writer Zhang, articles will be decrypted in detail 1) chat bot to solve three problems; 2) and the models they use.
Introduction:
Chatbot is very recently a word or an application of fire, not just the news media after the storm BOT concept, the Giants also spending huge resources on research and development, brush out the BOT-related paper on the arXiv is more the norm. Hype, hype, PR to the PR, have to say that one embarrassing fact is really difficult to find a really good bot on the market. Bot in accordance with related areas, is divided into an open field (open-domain) and specific task-oriented (task-oriented) bot. Open the domain to do very large, more like a platform that can do anything, no matter what your needs, it can be solved, sort of true AI means, while the bot task-oriented focus on doing one thing, booking, reservation, passport, and so on.
Said open domain bot, everyone contact up of is some answered very does not make sense of entertainment with bot, like many years Qian active in the big social website Shang of small yellow chicken, now market Shang active with many known as master has bot technology, in with depth learning solution bot technology of BOT company, are is this, solution can’t what actual problem, is can and everyone chat Shang two sentence, and many when answered are is irrelevant of, is funny.
Say task-oriented bot, market Shang up of is customer service robot, Bank or, electric business or, didn’t want to repeat sex to answered user of problem, on with a customer service robot to should, and not said effect how, development a specific task of BOT need fee many time, and late also to large of maintenance, because too more of hand crafted features was with to, whole bot of framework horizontal extended sex relative for poor, For one scene is basically just needs to develop a human costs are too high.
Ideal for BOT very plump, big scene at the company also really beautiful, but realised the BOT was poured a bucket of cold water. Expectations high, the greater the disappointment. If the media is always touted bot, as if the whole world tomorrow will be the bot, BOT are not beneficial development, pengsha will only lead to bubble, after the collapse, all the same.
Function powerful of, and open domain of BOT in short-term within is compared difficult achieved of, but if reduced expects, will bot not should do is a technology level of revolution, and should do interactive level of innovation is rational of attitude, bot as a entrance, may everyone are no longer need a carry carry of Terminal, only need found a can recognition identity, can networking of hardware, like side mirror, on can implementation many of task, set tickets, and buy things and so on and so on. BOT at this time is an action entry and perform various task behind the black box, we do not have to see the whole process, nor need to know what is, through simple language interaction, will be able to complete the complex task, the Terminal to do is feedback and receive input, process performed in the cloud, the bot cloud.
And all of this is the key to solve the task-oriented bot, with more data driven solutions to replace the traditional manual features and templates.
Dialogue is the last step, is part of the output. Under summary, there are four solutions:
Solution 1 directly from the context to generate dialogue, this recent paper is too much, especially after the seq2seq+attention framework has swept much of NLP tasks, dialog generates the benchmark model refresh over and over again. Dialogue generation problem, is defined as a build model based on a typical predict words based on the context, involves the question of sentence generation, evaluation can be a difficult problem.
Solution 2 of course paper is not a dialog build definitions for language modeling problems, but one of the next utterance selection issue, a multiple-selection of a problem, given a context, a given utterance candidate list, select one from the list as a response, of course, the difficulty of this type of problem is much smaller, is also very easy to evaluation, Data set ready to spend some time, and bad experience in the practical application.
Solution 3 rule-based or template-based,response end is actually filled in the form of a template, most things are given, only some specific value you need to fill. This type of solution is suitable for the task-oriented bot, too many artificial features and templates lead on difficult to transplant it to a different task.
Solution 4 query-based or example-based,response are from a database called the knowledge base, which contains a large, rich example, based on the user’s query, find the closest example, to return the corresponding response as the output. This type of solution is ideal for entertainment, comedy by the bot, the core technology is to find more data to enrich the knowledge base, to wash the base. But respnose from someone out there may be funny, but most will be irrelevant.
2、dialog state tracking(DST)
Some paper called DST belief trackers, this component is in fact the core of BOT, its role is to understand and capture the user intention or goal, only when you really know what users need, you can make the right action or response. On this section, there will be a Dialog State Tracking Challenge game. In General will be given a State-wide, to predict which state the user belongs to by context, what needs are need to query the weather or to locate the train tickets.
3、user modeling
BOT-oriented businesses are dealing and real user, if only a simple FAQ bot, answer some common questions you may not need this, but if it is more complex and delicate business, needed for user modeling, same problem, bot answered everyone’s response must be different, the reason is very simple. User modeling, involves more than just simple user basic information and some explicit user feedback, and more importantly users history of conversations, these implicit feedback. Before you fire up like recommendation system, everybody is quite satisfactory to sell stuff, but there are some smart analysis of users ‘ behavior, not only those who praise acts, more of those users inadvertently left the “clues” to know that the user is interested in what is potentially, then recommend what system do. For the modeling of user, is to be an individual BOT, each response that is generated has the user characteristics.
Corpus |
Large corpus is used to train the bot open Domain dialog model, data sources generally come from social networking sites. So far as the task-oriented bot, customer data is very small in size, it is also difficult to be data driven solution directly to one of the main reasons on the task-oriented bot.
[1] survey of BOT training corpus, interested students can be read about the survey.
From [13], the English’s corpus does more, Sina Weibo that the corpus laboratory of Huawei’s Noah’s Ark release [12]. From Twitter or have a bot on Twitter data, “conversational in nature” than the data generated from this chat room Ubuntu chat logs is more suitable for training response model, since the more natural pollution-free. [5] also used a large corpus, data from Baidu.
Model |
Research paper of the bot is too much, this is a very active area of research, segmentation is also very much in the direction of, according to the research question, then to introduce some models.
Seq2seq model
Now the most popular solution is the seq2seq+attention,encoder user query feed in, outputs a vector representation to represent the entire query, and then as a condition of the decoder and decoder is essentially a language model, step by step, to build response,[2] is such a programme, Google uses a ton of training parameters such a model, get a good bot.
Typical seq2seq there is a problem, that is likely to generate some “huh” response, that is something very safe,grammatical but no meaningful response, such as “I don’t know!” Or something like that. Because the traditional seq2seq in the decoding process are based on the MLE (Maximum Likelihood Estimate) as the objective function, which generated the most grammatical words, rather than the most useful, these safe sentence in large numbers to appear in the training corpus, after learning, inevitably always produce such a response, and the article [3] speech recognition experience for reference, In decoding with MMI (Maximum Mutual Information) as the objective function, improving the diversity of the response.
[4] that resemble the RNNLM of language models in high quality of generated words simply because without dealing with hidden feature or random noise in the utterance, thus generating next token (short term goal) and future tokens (long term goal) general effect.
When you build your each utterance, using four parts, encoder RNN, and context RNN and the latent variable and the decoder RNN, according to the order of input and output. Here the latent variable and IR LSI a little similar, latent indicates that we don’t know what they are, but probably represent a topic, or sentiment, is a reduction of the representation.
[5] proposed a technique called content method to generate a short response of the introducing.
Step 1 after the given query, predicting a topic keyword as a response, the topic of speech are noun, this keyword does not capture the complexity of meaning and grammar, but according to each word of the query to forecast a PMI (Pointwise Mutual Information) the highest term as a keyword.
Step 2 [5] model call Sequence To Backward and Forward Sequences, the first backward step, given a query, encoder Gets a context,decoder part of a given keyword, as the first word, then decoding, build this part of the equals keyword word in front of the section ; Is the next forward step, and is a typical seq2seq, represented by the encoder will query context, and then given backward and keyword as the first half of the decoder and continues decoding the latter part. Briefly describe the process as a whole:
[6] in question is the response the problem of inconsistency in the round of talks, the user identity (such as background information, user picture, age, and so on) taking into account the model, build a personalized seq2seq models for different user, as well as the same user on different generates a different style of response.
[6] model called Speaker Model, is a typical seq2seq model is different in the decoding section adds a speaker embedding, similar to the word embedding, except to say that users are modeling here. Because user information cannot be explicitly modeled, using a method for embedding, trained to be speaker vector, the following figure on the left is represented on speaker vector in a two-dimensional plane, with similar background information of the user will be very close, and the word vector a truth.
Reinforcement learning model
Enhanced learning to interactive problem-solving has a long history, but with the hype of AlphaGo, deepmind will enhance the learning back to the stage, combined with deep learning to solve some of the more difficult problems.
Enhance learning with a long term reward as the objective function, will make the training model can predict the response of higher quality, article [7] propose a model framework, and has the following capabilities:
1. custom integration developer the reward function, to achieve the goals.
2. generate a response later, can be used to describe the effect of response to subsequent phases.
Two bot in the dialog class when given an input message and bot1 generates 5 candidate response based on input, and proceed, because each has a 5 response input, with the increase in turn, exponential growth response, each round of dialogue, select 5 by sample, as this response.
Training in a large data set a good seq2seq as the initial value, using reinforcement learning to enhance the model’s ability to implement custom reward functions in order to achieve the expected results.
[7] the model can generate more rounds of dialogue, rather than prematurely into an infinite loop, generate dialogue and diversity is very good.
Task-oriented seq2seq model
Existing task-oriented bot is using rule-based, template-based or example-based, or integrated with, data driven solutions to extremely rare. Article [8] and [9] is to try the BOT on the individual parts of the technical depth to do and come up with a practical plan.
[8] first start from a familiar scene introduces how an experienced customer service with a new service, divided into four phases: Dolce & Gabbana iPhone 6 Plus Case
1. new customer service which “controls” are available, such as: information on how to find customers, how to determine their identity, and so on.
2. new good examples of imitative learning to obey the old customer service.
3. the new customer service trying to service customers, old customer service promptly corrected his error.
4. the old customer service an absentee, the new customer service customer service alone, continuous learning, and experience.
[8] model is designed in accordance with the procedures of the framework:
Developer offers a range of alternative actions, including response and some API functions, used to be called bot.
Experts offer a range of example dialogues using RNN learning.
Simulating user with a randomly generated query,bot response, correct expert.
BOT on-line services, and real customers to engage in a dialogue to improve bot service quality through feedback.
A complete workflow from the above diagram describes specific steps below:
Training is part of the quality of monitoring data to learn SL, RL with enhanced learning model received higher-quality results.
[9] the balance of advantages and disadvantages of two popular programs, presented a set of valuable, practical seq2seq solution.
A total of five components:
1、 Intent Network
The encoder part of the part can be understood as seq2seq would encode user input into a vector.
2、 Belief Trackers
Also known as Dialogue State Tracking (DST), are core components of the task-oriented bot. This Belief Trackers have the following roles:
Support various forms of natural language is mapped into a finite slot-value element in the collection, for the query in the database.
Tracking the State of BOT to avoid learning that no amount of data.
Use a weight tying strategy, you can greatly reduce the need for training data.
Extensible new component.
3、 Database Operator
Belief Trackers-input from the output of a database query, the probability distribution of the slot, the DB maximum input, query gets to the appropriate value.
4、 Policy Network
This component is like a glue, play the role of bond above the other three components. Input is the output of the above three components, the output is a vector.
5、 Generation Network
Last component is to generate model is essentially a language model, the input is the output of the Policy Network and the output is a generated response, after some processing steps can be returned to the user. Here, the treatment is to slot in the response, such as s.food and returned to the true value. This step and the article [8] step 101 will be specified on the restore value to the entity.
Completely solved with end-to-end task-oriented is impossible things, must be in a framework or system using the seq2seq solution to do this thing, [8] and [9] gives a lot of inspiration.
Knowledge Sources based model
Pure seq2seq can solve a lot of problems, but if for a specific task, in seq2seq on the basis of additional related knowledge sources will make many good results. Knowledge here can be sources of unstructured text, such as article [10] Ubuntu manpages, can also be a structured business data, such as articles [9] in the database, or it can be a source of data and business data to extract knowledge graph.
[10] the authors define a bot task as next utterance classification, a bit like question answering tasks, given a context and a response candidate list as an alternate answer, use context to choose the correct response from the candidate list. Contribution of this paper is based on the context, introduced the task relevant external expertise, and the knowledge base is structured.
Model is composed of three RNN encoder, one to encode context RNN, RNN to encode a response, there is an RNN to encode knowledge, and then combine to make predictions, to choose the most appropriate response. Model is called knowledge encoder. Because the DataSet using the Ubuntu technical support-related data sets, external resources will use Ubuntu manpages.
Context sensitive model
[11] the model is relatively simple, but means a lot of issues, history of information modeling for the bot is a great help in solving practical engineering applications, determines whether your bot can work. Author history context Word bag model, instead we use the RNN, context and user query is then passed through a simple FNN, get an output.
Evaluation |
BOT response evaluation is difficult, although BLEU can learn from the automatic evaluation of machine translation method to do, but not very good. Almost every paper is money to hire people to do manual evaluation, designing an evaluation mechanism to rate, human evaluation is more persuasive. Is especially true for practical application, users say is really good. Rather than simply holding their own, biased indicator, and several methods of comparing the bot or any other company, to explain himself.
Thinking |
Read the paper, and after the bot application engineer of communication, reflection, summarized as follows:
1, do you want to do bot? Popular is a theory is no easy to use bot on the market, to solve the problem of BOT needed a lot of technology and progress, may take a very long period of time, to do business with this thing now, is ridiculous. My personal view is that the bot to solve specific task, combining advanced technology, do some framework tool, not something so far, although it is not easy, but it’s very meaningful, solve the problem of vertical field of BOT, it is possible to solve the open domain problem of BOT. Is also not easily, raising the bar, real opportunities arise, the birth of some great technology companies.
2, open domain is task-oriented? If it were me, I would choose the latter, because the former is only a dream, a distant dream, more technical breakthroughs are needed. Task-oriented more specific, more practical, specific business, offer some solutions, there are already many companies do, although a generality or extensibility solution also did not appear, but a trend is a new generation the opportunity of companies doing bot.
3, task-oriented why the bot, which direction the force? End-to-end is a idealized of model, with depth learning model from large training data in the to “capture” some features, “intends collection” some function, although can get is good of effect, and using up does is convenient, but embarrassing on embarrassing in specific of task in the is took not to mass data of, data scale small has zhihou, purely of end-to-end on became very chicken has. In real-world scenarios, however, many enterprises have certain data or have a bot needs, so now is a mature solution for your specific business, to designing features,templates and rules, when the customer’s business changes, need to continue to maintain the existing bot system is very time-consuming. Real scene often involves a lot of structured business data in purely to directly generate response based on context, violence is possible, the article [8][9] are very enlightening solutions are given, use end-to-end in local, rather than a whole, coupled with technologies such as Information Extraction and Knowledge on Graph, achieving a highly available architecture, This should be a task-oriented bot development direction.
4, response generation and what should be on this? Response quality is good or bad, you need links to these features: (1) the user query, a user’s question, what users are asked in this round of dialogue, precise understanding of the user’s intent, this is crucial. (2) user modeling, user modeling, including basic information about users, and more importantly users the mining history conversation logs, this work is very hard, but at the same level, is a technology company to prove their technology a way to cool. Now logs mining is common, not everyone did well, and the logs are not set, structured index, but unstructured text logs, and began digging harder. Another point, paper types, user emotion and sentiment analysis is more NLP Research task, the user’s mood is directly related to the success of sales, if enough cows, factors to be considered will be enough analysis of the user are clear enough. History hung in the model is not a good idea, because history is growing problem will cause the model to capture info, the better approach might be to build things like user profile, history precipitated as a vector representation, or a knowledge graph to represent a user. With this bot, say sounding a little personalized bot. (3) knowledge external knowledge sources, when related to the specific business, business data is a kind of knowledge, how knowledge modeling to model, generate dialogue can be more professional and precise is also a very important issue. Bot is a general problem, not just hard on the system, and is modeled on the hardest.
5, I have always felt that life and will have to look at the issue, the world is not black and white, but rather somewhere in between successive values. Can’t say either as a open-domain Big Mac Bot or bot without a specific function, can not only see the existing bot is not mature, and BOT out of reach in the fantasy, dark this area, laugh at somebody else can get investment. Fight these senseless, really meaningful is dig deep in this area, identify pain points and difficult points of breaking one by one, continue to promote development in this area, rather than some of the streets to watch the man, it’s boring! Before the breakthrough in many fields, as if never see dawn, but not all of the many problems that cannot be solved in a few years is a Red Sea, streets are now? A generic bot for a long time is a difficult thing, but a high availability, extensibility, good bot solution there is to look forward, not overconfident, nor to belittle and get down to do it.
Lei Feng network Note: this authorized by ResysChina reprinted, for reprint, please contact the original author.
Past high profile flat coming from international big names, such as Samsung, Asus and so on, of course Apple in this regard is not fuel-efficient lights. But with mobile SoC development to the A15 ‘s age, the pattern seems to be changing that many unknown Tablet brands are beginning to “nuclear war”, some brands have more direct use of A15 core SoC, achieved industry-leading configuration.
Tegra 4
Tegra 4 from released has has been by widely concern, it using has highest 1.9GHz of four nuclear Cortex-A15 schema processor, 72 “nuclear” of GPU, performance of tough no capacity questioned, certainly limit it development of main reasons is power bad control, market Shang using it of equipment numbered, which NVIDIA home of Palm machine Shield more is directly added has thermal fan, visible its temperature of high.
Tablet from Shenzhen Hui Wei brand pioneered the first Tegra 4 tablet, and joined a number of unique features, claiming that run up to 36000. About Tegra 4 tablet, we have reviewed a Toshiba AT703, have recently been experienced the power of Tegra 4, but due to the AT703 of the resolution is too high, optimization is not very good, so the General performance. But this meeting via Pad can be seen but came prepared, the resolution should estimate the level of control in the 1080P or even 720P, it also makes Tegra 4 has more room to play. Meanwhile, Hui Wei Pad also acceded to a number of unique features, such as hyper-sensitive stylus, dual PureAudio technology.
Although we haven’t seen the meeting power Pad, but officials said Tutu 36000 properties of its powerful, relatively low resolution allows GPU workload reduction, better performance.
Exynos 5410
Exynos 5410 is so powerful a SoC in the Android camp, it uses four 1.6GHz Cortex-A15 and four 1.2GHz Cortex-A7 formed the “4+4” structure, machine Emperor Galaxy S4 version of Samsung’s active use (September 4 Galaxy Note III to replace it, of course). Local flat panel manufacturers long eight-core and sky have been launched using this chip plate. Small series of Huawei UltraStick press contact friends of sky S7801, it uses a Samsung Exynos 5410 processor, 7.85 1024*768 screen, 2G RAM, Max 32G ROM, camera front 200W, rear-500W/800W/1300W optional, 4000mAh battery. This low resolution can bring superior graphics performance, but the particles are relatively large screen. Here are some genuine real shot:
Exterior design, Samsung style, but working to rough a lot
The top two speakers
The side buttons
UltraStick related interfaces and Huawei at the bottom of the slot
UltraStick can insert SIM card
As can be seen from the diagram, Tablet design and Samsung’s tablet is similar, but the workmanship is far worse, Huawei, the highlight is the support the latest release of UltraStick Internet module by inserting the module and SIM card which turned WiFi Tablet 3G tablet, but it can also be called. In the course of a brief demo, xiaobian this tablet experience and international brand large gap, first in the workmanship of the product is not a rank, followed by the screen Panel is not used well, color and viewing angle is not ideal.
Huawei UltraStick briefly here, this is a Tablet, dedicated data card. As long as the plate with a special interface, you can insert the data card, data card insert SIM card WiFi Tablet 3G tablet, part tablet can also call functions.
Press Conference unveiled the Huawei E1220
You can insert a standard SIM card
Using proprietary data interface
Through data cable, UltraStick can also connect a computer
Dragon cool some of the leaked eight-core:
Tablet not only using Samsung chips, even the shape and Samsung tablet look like
From the figure, you can faintly see SoC Exynos series
Tablet resolution is 1280×800, using Samsung’s original screen, 2GB 16GB ROM and RAM configurations, system for Android 4.2.2. In addition, the Tablet has a rear-mounted front-5 million pixels and 13 million pixels, it can be considered flat panel pixel in the most.
Two Exynos 5410 Tablet seems to be using the same mold, because of the flat interior space relatively large, I believe Exynos 5410 play will be better than the Galaxy on the S4, not easily reduced frequency.
From the domestic route of the new tablet, cling to the low-end products are mainly used for Cortex-A7 architecture, and high-end products have been used by Cortex-A15 Exynos and Tegra 4 5410. Other configurations, is quite balanced, more than most mix 2G Ram,16G Rom, high megapixel cameras, but has a lot of product on the screen using 720P products, and color and viewing angle of the screen Panel is not good. In addition most of these flat-panel system using the native Android, coupled with not excellent optimization, will be in daily use have a lot of impact on the user experience. Home plate but also in design, screen material, as well as all aspects of UI work harder, just SoC, inadequate RAM stack makes a Tablet into the elite ranks. Of course, we also saw some positive aspects, such as Huawei UltraStick for domestic Tablet product in the network provides an excellent solution. With the appearance of Tegra 4 and Exynos 5410 tablet, we also believe that there will be more good SoC (long 800, Exynos, such as mycophenolate mofetil 5420) joined, will make the flat performance to the next level.
Related links:
Popular science: the four heavenly guardians Qualcomm “8064” know how many?
A15 lost? Xiao long 800 VS Exynos 5410 performance testing
July best equipment inventory: the onslaught of Xiao long 800lai
Tegra 4 how strong? Toshiba AT703 performance test
Four 2S, P6 nuclear Peterborough Z5 mini, millet Cross review
We had a taste of Android on Nexus 5 4.4 characteristics of, for the time being but “five-Port-au-Prince”, other transplant is our first phone version of early adopters. Millet 2 series (including millet, 2, 2S, 2A) has launched a transplanted native Android version 4.4, we of course also try Kitkat on millet. Flawless 6s case
In this paper, using a native Android phone 2 16G flush 4.4, brushing machine from “Autumn” autumn leaves blowing in the wind @ Ivan.
Flash method is very simple, only need to download the dedicated Recovery and brushing machine can be brushed into the. Brush into the previously prepared as shown in the figure four ZIP brushes packs, from top to bottom are the latest full MIUI 2S millet, Google service package, the package Android package, Android 4.4 4.4 brush machine-specific Recovery
Recovery in system upgrades page, click “install package”
Flush after Recovery, brush through it into the Android 4.4 brush Pack
Due to the millet has a dual system of setting, so the brush into the Android after 4.4 may still enter the MIUI, at this point we need to install a “dual-system switch” application because this application needs to obtain Root permissions, it is recommended users to brush into the development version of the early adopters of the MIUI, convenient system switch
Switching system, we can properly enter the native Android 4.4
Because uses the native Android system 4.4, so from the initiator and Nexus 5 we reviewed are nearly the same, millet 2S, of course there will be no virtual keys
2 s on Android 4.4 uses the source code to compile, on the second level of the menu icons are not Nexus 5 is so big, and icon styles Android 4.3 style
And Nexus to other 5 on almost exactly the same, here are two-drop bar
The Android version 4.4 may still start Google voice Now, of course, need to change system language into English
Android 4.4 of millet on the 2S remains a message
Setting interface
Android version 4.4, kernel version 3.4.0
Developer options will still be ART environment
Android 4.3, Adreno 320 and Adreno 330 supports OpenGL ES 3.0, millet 2S also support here, of course. But GPU drivers still V14, relatively old, Nexus 5 has to V53
Running and not much improve, but supports OpenGL ES3.0 can bring more to the game excellent graphics performance
This porting Android 4.4 millet 2 series were very fast, but from a practical point of view, MIUI V5 is the best partner of millet 2. Primary system has brought the operation experience of the relatively new though, and support for OpenGL ES3.0, but lost a lot of MIUI custom functionality, and optimization of the camera will be a lot worse, so it is suitable for users who want early adopters, are not suitable for long-term use.
Related links:
First taste of Kitkat Google Nexus 5 detailed evaluation
Four Smartphone high-end SoC horse race analysis
First Tegra4 millet 3 Edition detailed evaluation
Who is better? Mycophenolate Mofetil Dragon 800, Exynos version 5420 Note 3 showdown
Baqi dual quad core version Exynos5420 Note3 evaluation
A Travel Guide in China 