lecture stringclasses 575
values | image_name stringlengths 21 21 | source stringlengths 6 13.2k | target stringlengths 10 8.62k | image imagewidth (px) 480 1.28k |
|---|---|---|---|---|
CHI-F48F3 | CHI-F48F3-0081450.jpg | unfortunately, the Hci community is yet to give a lot of attention to this issue of responsibility. And has focused on trust fairness, which are also extremely important issues posed by by autonomous systems. And the few studies that actually address responsibility often use scenarios that are not realistic, such as th... | للأسف، مجتمع التفاعل بين الإنسان والحاسوب لسه ما أدّاش الاهتمام الكافي لقضية المسؤولية. وركز على الثقة والإنصاف، اللي هما برضه قضايا في غاية الأهمية اللي بتطرحها الأنظمة الذاتية. والقليل من الدراسات اللي بالفعل بتتناول المسؤولية غالباً بتستخدم سيناريوهات مش واقعية، زي مشكلة الترولي، وبتعامل المسؤولية واللوم كأنهم متراد... | |
CHI-F48F3 | CHI-F48F3-0100450.jpg | However, that is not really accurate. There's a lot of literature in the law, psychology, and philosophy that discuss this pluralistic concept of what responsibility is, right, and in this work, we actually focus on eight distinct notions that could apply to both humans and AI agents. | لكن ده مش دقيق فعلاً. في أدبيات كتير في القانون وعلم النفس والفلسفة بتناقش المفهوم التعددي للمسؤولية، صح؟ وفي العمل ده، إحنا بنركز فعلاً على تمن مفاهيم مختلفة ممكن تنطبق على البشر وعملاء الذكاء الاصطناعي. | |
CHI-F48F3 | CHI-F48F3-0131400.jpg | As a background, we actually use the Compas algorithm, right the? The system that has been used for for bail decisions in the Us. and that has received a lot of attention like. Like, by bias and fairness, literature in the past, and it basically posed the question of how to. lay people attribute responsibility to both ... | كمقدمة، استخدمنا خوارزمية كومباس، صح؟ النظام اللي تم استخدامه في قرارات الكفالة في الولايات المتحدة. واللي حصل على اهتمام كبير زي الانحياز والإنصاف، والأبحاث السابقة ناقشته، وهو في الأساس طرح سؤال عن إزاي الناس العاديين بينسبوا المسؤولية لأنظمة الذكاء الاصطناعي والبشر المشتركين في قرارات الكفالة. | |
CHI-F48F3 | CHI-F48F3-0162450.jpg | is that people attribute responsibility in three different clusters. And the 1st one is forward looking, That actually concerns things that have not yet occurred. The second one is this responsibility for explaining or justifying one's decisions, and the third one is backward-looking, which actually concerns things tha... | هو إن الناس بينسبوا المسؤولية لثلاث مجموعات مختلفة. وأول نوع هو النظر للأمام، وده بيتعلق بحاجات لسه ما حصلتش. والنوع التاني هو مسؤولية شرح أو تبرير قرارات الشخص، والنوع التالت هو النظر للخلف، وده بيتعلق بحاجات حصلت بالفعل في الماضي. كمان لاحظنا اختلافات واضحة، زي بين الذكاء الاصطناعي والبشر في التلات مجموعات دي من | |
CHI-F48F3 | CHI-F48F3-0172450.jpg | responsibility. First, we found that people expect both humans and AI advisors to explain their decisions, which highlights the importance of explainable AI. | المسؤولية. أولاً، لقينا إن الناس بيتوقعوا من البشر ومستشاري الذكاء الاصطناعي إنهم يشرحوا قراراتهم، وده بيبرز أهمية الذكاء الاصطناعي القابل للتفسير. | |
CHI-F48F3 | CHI-F48F3-0194450.jpg | On the other hand, we found that humans were attributed higher levels of this forward-looking notions of responsibility, and that they were attributed higher obligation, more authority, more tasks, and more skills For making assisting their decisions than their Ai systems | ومن ناحية تانية، لقينا إن البشر نسب ليهم مستويات أعلى من المفاهيم دي للنظر للأمام بالنسبة للمسؤولية، وإنهم نسب ليهم التزام أعلى، وسلطة أكتر، ومهام أكتر، ومهارات أكتر في اتخاذ القرارات والمساعدة فيها مقارنة بأنظمة الذكاء الاصطناعي | |
CHI-F48F3 | CHI-F48F3-0224450.jpg | counterparts, on the other hand. what we found is that humans and AI were blamed to a similar level. They also expected These two entities to compensate those harmed to the same extent, we explained these results based on previous work in cognitive science. and we also discussed what would it mean to blame an AI system... | وعلى الجانب الآخر، اللي لقيناه هو إن البشر والذكاء الاصطناعي تم لومهم على مستوى مشابه. كمان الناس توقعت إن الكيانين دول يعوضوا المتضررين بنفس الدرجة، وشرحنا النتائج دي بناءً على أبحاث سابقة في علم الإدراك. واتكلمنا كمان عن معنى لوم نظام ذكاء اصطناعي. فيه كمية كبيرة من الأدبيات، خصوصاً من الفلاسفة، بتناقش إذا كان ده ممك... | |
CHI-F48F3 | CHI-F48F3-0279450.jpg | Well, in summary, We actually stated how how late people attribute the responsibility to both humans and Ai agents in the context of their decision making. 1st, we found that people expect these two entities to explain their decisions, which that underlines the importance. Of explainable algorithmic decision making 2nd... | طيب، باختصار، احنا فعلاً وضحنا ازاي الناس بينسبوا المسؤولية للبشر وللوكلاء الذكاء الاصطناعي في سياق اتخاذ قراراتهم. أولاً، لاقينا إن الناس بيكون عندهم توقع إن الكيانين دول يوضحوا قراراتهم، وده بيأكد على أهمية اتخاذ القرارات اللي بتتميز بالشفافية. ثانياً، لقينا إن البشر بينسبلهم مستويات أعلى من المسؤولية الأخلاقية اللي ... | |
CHI-E64C0 | CHI-E64C0-0022450.jpg | Our paper, TapNet, is about the design, training, implementation, and application of a multitask learning CNN for off-screen mobile input. My name is Michael, and this is a collaboration work with Yang, Nasnee, Alex, and Xiumin. | ورقتنا، TapNet، بتتناول تصميم وتدريب وتنفيذ وتطبيق شبكة تعلم متعددة المهام CNN لإدخال الموبايل خارج الشاشة. أنا اسمي مايكل، وده عمل بالتعاون مع يانج، ناسني، أليكس، وزيومين. | |
CHI-E64C0 | CHI-E64C0-0061450.jpg | TapNet is a multitask network learned from coarse device data. It has multiple input branches, each of which accept data from one specific device. The multi-branch architecture enables the joint learning using coarse device data. TapNet is also a multitask network, and each output branch recognizes one tap property, in... | TapNet عبارة عن شبكة متعددة المهام متعلمة من بيانات الجهاز الخشنة. فيها فروع إدخال متعددة، وكل فرع بيستقبل بيانات من جهاز معيّن. العمارة متعددة الفروع تمكن التعلم المشترك باستخدام بيانات الجهاز الخشنة. TapNet برضه شبكة متعددة المهام، وكل فرع خرج بيعرف خاصية واحدة للضغط، بما يشمل اتجاه الضغط، أجزاء الإصبع، والموقع. العم... | |
CHI-E64C0 | CHI-E64C0-0091450.jpg | properties. And here is a demonstration of that about the tapnet results. The heatmap indicates the tap location. The color it denotes the tap direction, green for back, red for front and blue for side taps. The texture here shows. the finger parts, and the dotted pattern indicates the tapping using fingernail, and the... | والخصائص دي عليها عرض للنتائج من TapNet. خريطة الحرارة بتوضح موقع الضغط. اللون بيمثل اتجاه الضغط، الأخضر معناه للخلف، الأحمر للأمام، والأزرق للجوانب. هنا النسيج بيبين أجزاء الإصبع، والنمط المنقط بيعني الضغط باستخدام الظفر، والنمط الغير منقط بيعني الضغط بوسادة الإصبع. | |
CHI-E64C0 | CHI-E64C0-0107450.jpg | And here is a glance at the accuracy and sensitivity. TapNet achieved roughly one-tenth of the screen diagonal for tap location estimation. Gentle taps can be recognized. | وهنا لمحة عن الدقة والحساسية. TapNet حقق تقريبًا عشر الشاشة القطري لتقدير موقع الضغط. الضغطات الخفيفة ممكن تتعرف. | |
CHI-E64C0 | CHI-E64C0-0118450.jpg | We also include diverse non-tap motion in training, as such, TapNet can distinguish tap events from motion, such as tap release, phone shaking, etc. | كمان أضفنا حركات غير الضغط في التدريب، على هذا الأساس TapNet يقدر يفرق بين أحداث الضغط والحركة، زي تحرير الضغط، هز الموبايل، وغيرها. | |
CHI-E64C0 | CHI-E64C0-0170450.jpg | This is the TapNet pipeline. Its inputs include IMU signals and device vectors. The device vector here indicates the IMU install location, as well as the phone size. The core of the TapNet is a multi-layer convolutional neural network. It has multiple Convolutional layer Max pooling, and for each branch output branch, ... | دي بايبلاين TapNet. المدخلات بتاعتها بتشمل إشارات IMU ومتجهات الجهاز. متجه الجهاز هنا بيبين مكان تركيب IMU وكمان حجم التليفون. قلب TapNet هو شبكة عصبية متعددة الطبقات بالالتفاف. فيها طبقات كتير من الالتفاف مع التجميع الأقصى، ولكل فرع مخرجات، فيها طبقات متصلة بالكامل. وكل فرع مخصص لمهمة التعرف معينة، زي كشف حدث الطب، تص... | |
CHI-E64C0 | CHI-E64C0-0180450.jpg | property recognition branches, we use the diverse tapping action with annotation about the direction, location, and tap finger parts for training. | فروع التعرف على الخصائص، بنستخدم حركات الطب المختلفة مع التوضيح بالنسبة للاتجاه، المكان، وأجزاء إصبع الطب للتدريب. | |
CHI-E64C0 | CHI-E64C0-0191450.jpg | And for the Tap event detection branch, we use the both tap data as well as the non tap data. for training. | وبالنسبة لفرع كشف حدث الطب، بنستخدم بيانات الطب وبيانات غير خاصة بالطب للتدريب. | |
CHI-E64C0 | CHI-E64C0-0205450.jpg | We collect a one-person training dataset, meaning that the TapNet is trained on the data from only a single person. And then we collect a multi-person test set, | احنا بنجمع مجموعة بيانات تدريب لشخص واحد، يعني انه TapNet بيتدرب على بيانات من شخص واحد فقط. وبعد كده بنجمع مجموعة اختبار لعدة أشخاص، | |
CHI-E64C0 | CHI-E64C0-0208450.jpg | so our model is evaluated on multiple | فالموديل بتاعنا بيتم تقييمه على عدة أشخاص. | |
CHI-E64C0 | CHI-E64C0-0227450.jpg | persons. We use weighted average F1 score to measure the classification tasks. We use mean absolute error and R-squared to measure the accuracy of the regression tasks. TapNet can either perform on-par with or outperform the prior | الأشخاص. إحنا بنستخدم المتوسط المرجح لدرجة الـ F1 عشان نقيس مهام التصنيف. وبرضه بنستخدم متوسط الخطأ المطلق ومعامل التحديد (R-squared) عشان نقيس دقة مهام الانحدار. TapNet ممكن يا إما يقدم نفس الأداء أو يتفوق على الأداء السابق. | |
CHI-E64C0 | CHI-E64C0-0254450.jpg | arms. A close examination of the normalized convolution matrix of the 35-class tap location classification shows that neighboring region above or below the target has an index of offset 5. And the three parallel lines here along the diagonal indicate the TapNet either predicts correctly or to nearby regions. | الأذرع. فحص قريب لمصفوفة الالتفاف المعيارية لتصنيف موقع النقر بـ 35 فئة بيبين إن المنطقة المجاورة فوق أو تحت الهدف عندها مؤشر انحراف قيمته 5. والتلات خطوط المتوازية هنا على طول القطر بتوضح إن TapNet يا إما بتتوقع بشكل صحيح أو للمناطق القريبة. | |
CHI-E64C0 | CHI-E64C0-0302450.jpg | There are a few takeaway messages from this paper. For E.g. the training strategy using one person data can be effective for gesture recognition systems. Multi-task learning can be more data efficient in our tasks, in particular when there is only limited Amount of training, data and cost device training with the phone... | فيه شوية نقاط مستفادة من الورقة دي. على سبيل المثال، استراتيجية التدريب باستخدام بيانات شخص واحد ممكن تكون فعالة في أنظمة التعرف على الإيماءات. التعلم متعدد المهام ممكن يكون أكتر كفاءة في استخدام البيانات في المهام بتاعتنا، خصوصاً لما يكون فيه كمية محدودة من التدريب والبيانات وتكاليف تدريب الجهاز بالتليفون. أشكال الأجه... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.