48 skills found · Page 1 of 2
codebasics / Data Structures Algorithms PythonThis tutorial playlist covers data structures and algorithms in python. Every tutorial has theory behind data structure or an algorithm, BIG O Complexity analysis and exercises that you can practice on.
farlleyferreira / Ms Fastapi TemplateThis project was built as a result of a deepening of the studies discussed on the blog farlley.com with a greater focus on Domain Driven Design (DDD) architecture. In this work you will find a simple template for creating microservices, as well as a use case (which will still be implemented according to the Roadmap found in this same document) applying the theories and studies presented in the books Implementing Domain-Driven Design - Vaughn Vernon, Domain-Driven Design: Attacking the Complexities at the Heart of Software - Eric Evans, and Building Microservices: Designing Fine-Grained Systems - Sam Newman, the bibliography consulted will be added to the end of this document in accordance with the application roadmap. The template was written for the python language in its version 3.9.0, using the FastApi framework.
SamuelSchlesinger / Complexity ResourcesResources and notes on computational complexity theory
iepathos / DebtmapTechnical debt and risk analyzer that predicts bug hotspots by combining cognitive complexity, pattern recognition, coverage gaps, information theory, and git history.
marcincuber / Modal LogicFinal Year Masters Project: modal logic solver tableaux
storopoli / Graphs ComplexityGraph Theory and Computational Complexity
Mariusmarten / Complexity ResourcesComplex Systems, Complexity, Chaos, Emergence, Network Theory, Nonlinear, Graph theory
sainathiyer / Hardware Implementation Of Polar Code For 5G Application On FPGAPolar codes are error correction codes developed by Erdal Arikan which achieves channel capacity and its reduced complexity makes it more attractive and successful. The discovery of polar code is standing as a milestone in coding theory and lot of researches are carrying out on this topic. Varieties of schemes have been proposed over these years for the generation, encoding and decoding of polar codes. The important area of research is the encoder and decoder section and most widely used one is successive cancellation decoder. Reduced complexity is the most attractive feature with an overall encoding and decoding complexity of O (NlogN) for a block size of N, which leads to its great success. Polar code concept is used in the most promising 5G technology. In this project a complete implementation of the polar Encoder and decoder is discussed with its MATLAB and Verilog implementations.
SE-Design / FAQ.mdNetSaver Pro ======== Please scroll down if you want to ask a question, request a feature or report a bug. Frequently Asked Questions (FAQ) -------------------------------- <a name="FAQ0"></a> **(0) How do I use NetSaver Pro?** * Enable the firewall using the switch in the action bar * Allow/deny Wi-Fi/mobile internet access using the icons along the right side of the application list You can use the settings menu to change from blacklist mode (allow all in *Settings* but block unwanted applications in list) to whitelist mode (block all in *Settings* but allow favorite applications in list). * Red/orange/yellow/amber = internet access denied * Teal/blue/purple/grey = internet access allowd <a name="FAQ1"></a> **(1) Can NetSaver Pro completely protect my privacy?** No - nothing can completely protect your privacy. NetSaver Pro will do its best, but it is limited by the fact it must use the VPN service. This is the trade-off required to make a firewall which does not require root access. The firewall can only start when Android "allows" it to start, so it will not offer protection during early boot-up (although your network may not be loaded at that time). It will, however, be much better than nothing, especially if you are not rebooting often. If you want to protect yourself more, you can (at least in theory) disable Wi-Fi and mobile data before rebooting, and only enable them on reboot, after the firewall service has started (and the small key icon is visible in the status bar). Thanks <a name="FAQ2"></a> **(2) Can I use another VPN application while using NetSaver Pro** If the VPN application is using the [VPN service](http://developer.android.com/reference/android/net/VpnService.html), then no, because NetSaver Pro needs to use this service. Android allows only one application at a time to use this service. <a name="FAQ3"></a> **(3) Can I use NetSaver Pro on any Android version?** No, the minimum required Android version is 4.0 (Lollipop) because NetSaver Pro uses the [addDisallowedApplication](http://developer.android.com/reference/android/net/VpnService.Builder.html#addDisallowedApplication(java.lang.String)) method. <a name="FAQ4"></a> **(4) Will NetSaver Pro use extra battery power?** If you didn't enable IP filtering, probably not. However, the network speed graph notification will use extra battery power. This is why the notification is shown only when the screen is on. You can decrease the update frequency using the settings to reduce the battery usage. <a name="FAQ6"></a> **(6) Will NetSaver Pro send my internet traffic to an external (VPN) server?** No, depending on the mode of operation basically one of two things will happen with your internet traffic: * When IP filtering is disabled, blocked internet traffic will be routed into the local VPN which will operate as sinkhole (in effect dropping all blocked traffic) * When IP filtering is enabled, both blocked and allowed internet traffic will be routed into the local VPN and only allowed traffic will be forwarded to the intended destination (so not to a VPN server) The [Android VPN service](http://developer.android.com/reference/android/net/VpnService.html) is being used to locally route all internet traffic to NetGuard so no root is required to build a firewall application. NetSaver Pro is unlike all other no-root firewalls applications. <a name="FAQ7"></a> **(7) Why are applications without internet permission shown?** Internet permission can be granted with each application update without user consent. By showing all applications, NetGuard allows you to control internet access even *before* such an update occurs. <a name="FAQ8"></a> **(8) What do I need to enable for the Google Play™ store app to work?** You need 3 packages (applications) enabled (use search in NetGuard to find them quickly): * com.android.vending (Play store) * com.google.android.gms (Play services) * com.android.providers.downloads (Download manager) Since the Google Play™ store app has a tendency to check for updates or even download them all by itself (even if no account is associated), one can keep it in check by enabling "*Allow when device in use*" for all 3 of these packages. Click on the down arrow on the left side of an application name and check that option, but leave the network icons set to red (hence blocked).The little human icon will appear for those packages. Note that NetSaver Pro does not require any Google service to be installed. <a name="FAQ9"></a> **(9) Why is the VPN service being restarted?** The VPN service will be restarted when you turn the screen on or off and when connectivity changes (Wi-Fi, mobile) to apply the rules with the conditions '*Allow when screen is on*' and '*Block when roaming*'. See [here](http://forum.xda-developers.com/showpost.php?p=65723629&postcount=1788) for more details. <a name="FAQ10"></a> **(10) Will you provide a Tasker plug-in?** If disabling NetSaver Pro is allowed to Tasker, any application can disabled NetSaver Pro too. Allowing to disable a security application from other applications is not a good idea. <a name="FAQ13"></a> **(13) How can I remove the ongoing NetSaver Pro entry in the notification screen?** * Long click the NetSaver Pro notification * Tap the 'i' icon * Depending on your device and/or ROMs manufacturer software customizations, you can be directed to either: * the **App Info** screen and you can uncheck '*Show notifications*' and agree to the next dialog * the **App Notifications** screen and you can toggle the '*Block*' slider to on Note that, whether or not you get a dialog warning to agree upon, this operation will disable any information or warning notifications from NetSaver Pro as well, like the new application installed notification. To read about the need for the notification in the first place, see [question 24](#FAQ24). Some Android versions display an additional notification, which might include a key icon. This notification can unfortunately not be removed. <a name="FAQ14"></a> **(14) Why can't I select OK to approve the VPN connection request?** There might be another (invisible) application on top of the VPN connection request dialog. Some known (screen dimming) applications which can cause this are *Lux Brightness*, *Night Mode* and *Twilight*. To avoid this problem, at least temporary, close all applications and/or services which may be running in the background. <a name="FAQ15"></a> **(15) Why won't you support the F-Droid builds?** Because F-Droid doesn't support reproducible builds. Read [here](https://blog.torproject.org/blog/deterministic-builds-part-one-cyberwar-and-global-compromise) why this is important. Another reason is that F-Droid builds are more often than not outdated, leaving users with an old version with known bugs. <a name="FAQ16"></a> **(16) Why are some applications shown dimmed?** Disabled applications and applications without internet permission are shown dimmed. <a name="FAQ17"></a> **(17) Why is NetSaver Pro so much memory?** It isn't, NetSaver Pro doesn't allocate any memory, except a little for displaying the user interface elements. It appeared that on some Android variants the Google Play™ store app connection, using almost 150 MB and needed for in-app donations, is incorrectly attributed to NetSaver Pro instead to the Google Play™ store app. <a name="FAQ18"></a> **(18) Why can't I findNetSaver Pro in the Google Play™ store app?** NetSaver Pro requires at least Android 4.0, so it is not available in the Google Play™ store app for devices running older Android versions. <a name="FAQ19"></a> **(19) Why does aplication XYZ still have internet access?** If you block internet access for an application, there is no way around it. However, applications could access the internet through other applications. Google Play services is handling push messages for most applications for example. You can prevent this by blocking internet access for the other application as well. Note that some applications keep trying to access the internet, which is done by sending a connection request packet. This packet goes into the VPN sinkhole when internet access for the application is blocked. This packet consists of less than 100 bytes and is counted by Android as outgoing traffic and will be visible in the speed graph notification as well. <a name="FAQ20"></a> **(20) Can I Greenify/hibernate NetGuard?** No. [Greenifying](https://play.google.com/store/apps/details?id=com.oasisfeng.greenify) or otherwise hibernating NetGuard will result in rules not being applied when connectivity changes from Wi-Fi/mobile, screen on/off and roaming/not roaming. <a name="FAQ21"></a> **(21) Does doze mode affect NNetSaver Pro?** I am not sure, because the [doze mode documentation](http://developer.android.com/training/monitoring-device-state/doze-standby.html) is not clear if the [Android VPN service](http://developer.android.com/reference/android/net/VpnService.html) will be affected. To be sure you can disable battery optimizations for NetSaver Pro manually like this: ``` Android settings > Battery > three dot menu > Battery optimizations > Dropdown > All apps > NetSaver Pro> Don't optimize > Done ``` This cannot be done from the application, because according to Google NetSaver Pro is [not an application type allowed to do this](http://developer.android.com/training/monitoring-device-state/doze-standby.html#whitelisting-cases). <a name="FAQ22"></a> **(22) Can I tether / use Wi-Fi calling while using NetGuard?** Yes, but this needs to be enabled in the settings. If it works depends on your Android version, because some Android versions have a bug preventing tethering and the VPN service to work together. Some devices hibernate Wi-Fi preventing tethering to work when the screen is off. This behavior can be disabled in the Android enhanced/advanced Wi-Fi settings. <a name="FAQ24"></a> **(24) Can you remove the notification from the status bar?** Android can kill background services at any time. This can only be prevented by turning a background service into a foreground service. Android requires an ongoing notification for all foreground services to make you aware of potential battery usage (see [question 4](#FAQ4)). So, the notification cannot be removed without causing instability. However, the notification is being marked as low priority, which should result in moving it to the bottom of the list. The key icon and/or the VPN running notification, which is shown by Android and not by NetGuard, can unfortunately not be removed. The [Google documentation](http://developer.android.com/reference/android/net/VpnService.html) says: "*A system-managed notification is shown during the lifetime of a VPN connection*". <a name="FAQ25"></a> **(25) Can you add a 'select all'?** There is no need for a select all function, because you can switch from black list to white list mode using the settings. See also [question 0](#FAQ0). <a name="FAQ27"></a> **(27) How do I read the blocked traffic log?** The columns have the following meaning: 1. Time (tap on a log entry to see the date) 1. Application icon (tap on a log entry to see the application name) 1. Application UID 1. Wi-Fi / mobile connection, green=allowed, red=blocked 1. Interactive state (screen on or off) 1. Protocol (see below) and packet flags (see below) 1. Source and destination port (tap on a log entry to lookup a destination port) 1. Source and destination IPv4 or IPv6 address (tap on a log entry to lookup a destination IP address) 1. Organization name owning the IP address (need to be enabled through the menu) Protocols: * ICMP * IGMP * ESP (IPSec) * TCP * UDP * Number = one of the protocols in [this list](https://en.wikipedia.org/wiki/List_of_IP_protocol_numbers) * 4 = IPv4 * 6 = IPv6 Packet flags: * S = SYN * A = ACK * P = PSH * F = FIN * R = RST For a detailed explanation see [here](https://en.wikipedia.org/wiki/Transmission_Control_Protocol). <a name="FAQ28"></a> **(28) Why is Google connectivity services allowed internet access by default?** The Google connectivity services system application checks if the current network is really connected to the internet. This is probably done by briefly connecting to some Google server. If this is not the case, there will be an '!' in the Wi-Fi or mobile icon in the system status bar. Recent Android versions seem not to switch connectivity from mobile to Wi-Fi when the Wi-Fi network is not really connected, even though there is a connection to the Wi-Fi network (or the other way around). On Android 6.0 and later you might get a notification asking you if you want to keep this connection on or not. To prevent a bad user experience there is a predefined rule to default allow the Google connectivity services. <a name="FAQ29"></a> **(29) Why do I get 'The item you requested is not available for purchase'?** You can only purchase pro feature when you installed NetSaver Pro from the Play store. <a name="FAQ30"></a> **(30) Can I also run AFWall+ on the same device?** Unless you are just testing NetSaver Pro, there is no current reason to use them both, since they cover the same function (firewall), although with different base needs (AFWall+ needs a rooted device) and ways of doing their thing (AFWall+ uses iptables). Also you need to keep per applicaton access rules _always_ in sync, else the application will not be able to access the network, hence bringing another level of complexity when setting and assuring things work out. Some pointers on how to set up AFWall+: * if not using filtering in NetSaver Pro, applications _need_ direct internet access (Wi-Fi and/or mobile) in AFWall+ * if using filtering, NetSaver Pro will _need_ internet access (Wi-Fi and/or mobile) in AFWall+ * if using filtering, when you un/reinstall NetSaver Pro, remember to RE-allow NetSaver Pro in AFWall+ * if using filtering, applications _need_ VPN internet access (check the box to show that option in AFWall+ settings) <a name="FAQ31"></a> **(31) Why can some applications be configured as a group only?** For a lot of purposes, including network access, Android groups applications on UID and not on package/application name. Especially system applications often have the same UID, despite having a different package and application name, these are set up like this by the ROM manufacturer at build time. These applications can only be allowed/blocked access to the internet as a group. <a name="FAQ32"></a> **(32) Why is the battery/network usage of NetSaver Pro so high?** This is because Android contributes battery and network usage which is normally contributed to other applications to NetSaver Prod in IP filtering mode. The total battery usage is slightly higher when IP filtering mode is enabled. IP filtering mode is always enabled on Android version before 5.0 and optionally enabled on later Android versions. <a name="FAQ33"></a> **(33) Can you add profiles?** Profiles are inconvenient because they need to be operated manually. Conditions like '*When screen is on*' are on the other hand convenient because they work automatic. Therefore profiles will not be added, but you are welcome to propose new conditions, however they need to be generally usable to be included. As a workaround you can use the export/import function to apply specific settings in specific circumstances. <a name="FAQ34"></a> **(34) Can you add the condition 'when on foreground'?** Recent Android versions do not allow an application to query if other applications are in the foreground or background anymore, so this cannot be added. You can use the condition '*when screen is on*' instead. <a name="FAQ35"></a> **(35) Why does the VPN not start?** NetSaver Pro "asks" Android to start the local VPN service, but some Android versions contain a bug which prevents the VPN from starting (automatically). Sometimes this is caused by updating NetSaver Pro. Unfortunately this cannot be fixed from NetSaver Pro. What you can try is to restart your device and/or revoke the VPN permissions from NetSaver Pro using the Android settings. Sometimes it helps to uninstall and install NetSaver Pro again (be sure to export your settings first). <a name="FAQ36"></a> **(36) Can you add PIN or password protection?** Since turning off the VPN service using the Android settings cannot be prevented, there is little use in adding PIN or password protection. <a name="FAQ37"></a> **(37) Why are the pro features so expensive?** The right question is "*why are there so many taxes and fees*": * VAT: 25% (depending on your country) * Google fee: 30% * Income tax: 50% So, what is left for the developer is just a fraction of what you pay. Despite NetSaver Pro being *really* a lot of work, only some of the convenience and advanced features are paid, which means that NetSaver Pro is basically free to use. Also note that most free applications will appear not to be sustainable in the end, whereas NetSaver Pro is properly maintained and supported. <br />
dia2018 / What Is The Difference Between AI And Machine LearningArtificial Intelligence and Machine Learning have empowered our lives to a large extent. The number of advancements made in this space has revolutionized our society and continue making society a better place to live in. In terms of perception, both Artificial Intelligence and Machine Learning are often used in the same context which leads to confusion. AI is the concept in which machine makes smart decisions whereas Machine Learning is a sub-field of AI which makes decisions while learning patterns from the input data. In this blog, we would dissect each term and understand how Artificial Intelligence and Machine Learning are related to each other. What is Artificial Intelligence? The term Artificial Intelligence was recognized first in the year 1956 by John Mccarthy in an AI conference. In layman terms, Artificial Intelligence is about creating intelligent machines which could perform human-like actions. AI is not a modern-day phenomenon. In fact, it has been around since the advent of computers. The only thing that has changed is how we perceive AI and define its applications in the present world. The exponential growth of AI in the last decade or so has affected every sphere of our lives. Starting from a simple google search which gives the best results of a query to the creation of Siri or Alexa, one of the significant breakthroughs of the 21st century is Artificial Intelligence. The Four types of Artificial Intelligence are:- Reactive AI – This type of AI lacks historical data to perform actions, and completely reacts to a certain action taken at the moment. It works on the principle of Deep Reinforcement learning where a prize is awarded for any successful action and penalized vice versa. Google’s AlphaGo defeated experts in Go using this approach. Limited Memory – In the case of the limited memory, the past data is kept on adding to the memory. For example, in the case of selecting the best restaurant, the past locations would be taken into account and would be suggested accordingly. Theory of Mind – Such type of AI is yet to be built as it involves dealing with human emotions, and psychology. Face and gesture detection comes close but nothing advanced enough to understand human emotions. Self-Aware – This is the future advancement of AI which could configure self-representations. The machines could be conscious, and super-intelligent. Two of the most common usage of AI is in the field of Computer Vision, and Natural Language Processing. Computer Vision is the study of identifying objects such as Face Recognition, Real-time object detection, and so on. Detection of such movements could go a long way in analyzing the sentiments conveyed by a human being. Natural Language Processing, on the other hand, deals with textual data to extract insights or sentiments from it. From ChatBot Development to Speech Recognition like Amazon’s Alexa or Apple’s Siri all uses Natural Language to extract relevant meaning from the data. It is one of the widely popular fields of AI which has found its usefulness in every organization. One other application of AI which has gained popularity in recent times is the self-driving cars. It uses reinforcement learning technique to learn its best moves and identify the restrictions or blockage in front of the road. Many automobile companies are gradually adopting the concept of self-driving cars. What is Machine Learning? Machine Learning is a state-of-the-art subset of Artificial Intelligence which let machines learn from past data, and make accurate predictions. Machine Learning has been around for decades, and the first ML application that got popular was the Email Spam Filter Classification. The system is trained with a set of emails labeled as ‘spam’ and ‘not spam’ known as the training instance. Then a new set of unknown emails is fed to the trained system which then categorizes it as ‘spam’ or ‘not spam.’ All these predictions are made by a certain group of Regression, and Classification algorithms like – Linear Regression, Logistic Regression, Decision Tree, Random Forest, XGBoost, and so on. The usability of these algorithms varies based on the problem statement and the data set in operation. Along with these basic algorithms, a sub-field of Machine Learning which has gained immense popularity in recent times is Deep Learning. However, Deep Learning requires enormous computational power and works best with a massive amount of data. It uses neural networks whose architecture is similar to the human brain. Machine Learning could be subdivided into three categories – Supervised Learning – In supervised learning problems, both the input feature and the corresponding target variable is present in the dataset. Unsupervised Learning – The dataset is not labeled in an unsupervised learning problem i.e., only the input features are present, but not the target variable. The algorithms need to find out the separate clusters in the dataset based on certain patterns. Reinforcement Learning – In this type of problems, the learner is rewarded with a prize for every correct move, and penalized for every incorrect move. The application of Machine Learning is diversified in various domains like Banking, Healthcare, Retail, etc. One of the use cases in the banking industry is predicting the probability of credit loan default by a borrower given its past transactions, credit history, debt ratio, annual income, and so on. In Healthcare, Machine Learning is often been used to predict patient’s stay in the hospital, the likelihood of occurrence of a disease, identifying abnormal patterns in the cell, etc. Many software companies have incorporated Machine Learning in their workflow to steadfast the process of testing. Various manual, repetitive tasks are being replaced by machine learning models. Comparison Between AI and Machine Learning Machine Learning is the subset of Artificial Intelligence which has taken the advancement in AI to a whole new level. The thought behind letting the computer learn from themselves and voluminous data that are getting generated from various sources in the present world has led to the emergence of Machine Learning. In Machine Learning, the concept of neural networks plays a significant role in allowing the system to learn from themselves as well as maintaining its speed, and accuracy. The group of neural nets lets a model rectifying its prior decision and make a more accurate prediction next time. Artificial Intelligence is about acquiring knowledge and applying them to ensure success instead of accuracy. It makes the computer intelligent to make smart decisions on its own akin to the decisions made by a human being. The more complex the problem is, the better it is for AI to solve the complexity. On the other hand, Machine Learning is mostly about acquiring knowledge and maintaining better accuracy instead of success. The primary aim is to learn from the data to automate specific tasks. The possibilities around Machine Learning and Neural Networks are endless. A set of sentiments could be understood from raw text. A machine learning application could also listen to music, and even play a piece of appropriate music based on a person’s mood. NLP, a field of AI which has made some ground-breaking innovations in recent years uses Machine Learning to understand the nuances in natural language and learn to respond accordingly. Different sectors like banking, healthcare, manufacturing, etc., are reaping the benefits of Artificial Intelligence, particularly Machine Learning. Several tedious tasks are getting automated through ML which saves both time and money. Machine Learning has been sold these days consistently by marketers even before it has reached its full potential. AI could be seen as something of the old by the marketers who believe Machine Learning is the Holy Grail in the field of analytics. The future is not far when we would see human-like AI. The rapid advancement in technology has taken us closer than ever before to inevitability. The recent progress in the working AI is much down to how Machine Learning operates. Both Artificial Intelligence and Machine Learning has its own business applications and its usage is completely dependent on the requirements of an organization. AI is an age-old concept with Machine Learning picking up the pace in recent times. Companies like TCS, Infosys are yet to unleash the full potential of Machine Learning and trying to incorporate ML in their applications to keep pace with the rapidly growing Analytics space. Conclusion The hype around Artificial Intelligence and Machine Learning are such that various companies and even individuals want to master the skills without even knowing the difference between the two. Often both the terms are misused in the same context. To master Machine Learning, one needs to have a natural intuition about the data, ask the right questions, and find out the correct algorithms to use to build a model. It often doesn’t requiem how computational capacity. On the other hand, AI is about building intelligent systems which require advanced tools and techniques and often used in big companies like Google, Facebook, etc. There is a whole host of resources to master Machine Learning and AI. The Data Science blogs of Dimensionless is a good place to start with. Also, There are Online Data Science Courses which cover the various nitty gritty of Machine Learning.
Amey-Thakur / THEORY OF COMPUTER SCIENCECSC504: Theory of Computer Science [TCS] | TE Semester V | Computer Engineering
GiggleLiu / ProblemReductions.jlReduction between computational hard problems.
FuRuF-11 / AITA repository to introduce the algorithmic information theory. You could learn what is Kolmogorov complexity and why it is important here.
yhtang123 / Intelligent High Efficiency Energy Conversion SystemAbstract- As yet, the efficiency optimization of the power electronic converters needs to rely on its circuit model, while an inaccurate model cannot represent the correct operation behavior of the converter. Therefore, the accuracy of the power electronic converter model is of great importance for the efficiency optimization. However, the existing modeling methods cannot provide accurately model for the power electronic converters, since the parasitic parameters of its structure are closely related to the components and their layout, and the device structure size. Moreover, due to the power electronic converters usually contain many switching devices and their operating conditions are very complicated in practical application, thus many variable parameters need to be taken into account in the efficiency optimization process, which aggravates the computational complexity of optimization procedure for the efficiency. Based on the analysis mentioned above, the existing methods cannot provide the optimal efficiency optimization modulation strategy for the power electronic converters. Although, the artificial intelligence (AI) is powerful for solving the optimization and decision-making problems of difficult modeling and high-dimensional complex systems, its applications in the power electronic converters efficiency optimization are still being developed. Inspired by the successful application of the robotic chemist, and the classic games, here, we present an AI aided efficiency optimization engineer for the first time, which can train online to search for improved the operation efficiency for the dual active bridge (DAB) converter without prior knowledge about their circuit model. The engineer operated autonomously around the clock in a practical circuit platform about 71 hours, performing 120, 000 consecutive experiments within a six-variable experimental space, driven by the deep deterministic policy gradient (DDPG) algorithm. This autonomous exploring approach found an optimized modulation strategy, which can greatly improve the efficiency under entire continuous operation range compares to existing methods, especially under light load conditions. This online optimization approach can be deployed in the conventional power electronic converters for a range of operation performance optimization problems beyond the DAB converter. Our study created a novel idea and expanded the frontier theory for the power electronics automatic optimization.
ShrohanMohapatra / Matrix Multiply QuadraticThis source code (in Python) is a preliminary implementation of my quadratic-time positive integer matrix multiplication.
OpenWeavers / OpenanalysisAn implementation-neutral algorithm analysis and visualization library
mikinty / P Vs NPNotable P vs. NP proof attempts.
malena1906 / Pruning Weights With Biobjective Optimization KerasOverparameterization and overfitting are common concerns when designing and training deep neural networks. Network pruning is an effective strategy used to reduce or limit the network complexity, but often suffers from time and computational intensive procedures to identify the most important connections and best performing hyperparameters. We suggest a pruning strategy which is completely integrated in the training process and which requires only marginal extra computational cost. The method relies on unstructured weight pruning which is re-interpreted in a multiobjective learning approach. A batchwise Pruning strategy is selected to be compared using different optimization methods, of which one is a multiobjective optimization algorithm. As it takes over the choice of the weighting of the objective functions, it has a great advantage in terms of reducing the time consuming hyperparameter search each neural network training suffers from. Without any a priori training, post training, or parameter fine tuning we achieve highly reductions of the dense layers of two commonly used convolution neural networks (CNNs) resulting in only a marginal loss of performance. Our results empirically demonstrate that dense layers are overparameterized as with reducing up to 98 % of its edges they provide almost the same results. We contradict the theory that retraining after pruning neural networks is of great importance and opens new insights into the usage of multiobjective optimization techniques in machine learning algorithms in a Keras framework. The Stochastic Multi Gradient Descent Algorithm implementation in Python3 is for usage with Keras and adopted from paper of S. Liu and L. N. Vicente: "The stochastic multi-gradient algorithm for multi-objective optimization and its application to supervised machine learning". It is combined with weight pruning strategies to reduce network complexity and inference time.
0x0f0f0f / Computability Complexity Cheat SheetsTeX code for Computability Theory and Complexity Theory Cheat Sheets
MattiaOldani / Informatica TeoricaInformatica teorica
