ࡱ> g 6kbjbj .>r\r\6cbbbbbvvvv|vTp6666STTTTTT$.VX%Tb%Tbb66:T+++b6b6Rx+S++:+0,06B3X'#W0 sRPT0Ta0RdY%!dY00dYb3+%T%T-&TdY> (: Artificial Intelligence CSE 5290/4301, Spring 2020 Instructor: Debasis Mitra, Ph.D. Office: Harris 325 E-mail:  HYPERLINK "mailto:dmitra@zach. fit.edu" dmitra at cs.fit. edu Class Home Page:  HYPERLINK "http://www.cs.fit.edu/~dmitra/ArtInt/" http://www.cs.fit.edu/~dmitra/ArtInt/ Office Hours: MT 2-4 pm (or by appointment) Grading plan: Graduate stds: Quizzes (9, worst dropped): 15%, Coding exercise: 5%, Exams (other than 3-pointers) 40%, Project: 40% Undergraduate stds: Quizzes (~9, worst dropped): 20%, Coding exercises: 30%, Exams (other than 3-pointers) 50% Class TR 8:00-9:15pm Olin Eng 137  SYLLABUS FOR AI SPRING 2020 FINAL EXAM. Materials under *---* are excluded. 3.3-3.4.5 means 3.3 through 3.4.5 SEARCH, up to IDA* search, but not SMA*, Adversarial search (min-max) & alpha-beta pruning is included Up to materials on my slide Constrained Optimization Ch 4-4.2 Ch 5-5.3 CONSTRAINTS Ch 6-6.2.5, 6.3-6.3.2, 6.5 (*use of symmetries on p226*) Temporal reasoning: My slides LOGIC Ch 7.3-7.5.2, *completeness of resolution*, 7.5.3-7.6.2, *walksat algorithm*, Ch 8.2-8.3 Ch 9-9.2.2, 9.3-9.3.2, 9.4.1, 9.5-9.5.3 PROBABILISTIC REASONING Ch 13.2-end Ch 14-14.3, 14.7.2-3 LEARNING Decision Tree 18 18.3.4 Evaluation 18.4-Model Selection 18.4.1 Regularization 18.4.3 Learning Theory 18.5.0 only Regression 18.6 18.6.2 Classification 18.6.3 18.6.4 Neural Network 18.7 18.7.4 (exclude exotic varieties of NN on my slides) Non-parametric models 18.8 18.8.4 SVM basics 18.9 Clustering basics (from my slides) ETHICS My slides Florida Tech Academic Calendar: https://www.fit.edu/registrar/academic-calendar/ Detailed activities in Fall 2019 mapped on to the dates in Spring 2020, for now. This table will act as our continuously developing plan for this semester: ~ 28 meetings Lectures planned: Search 4, Constraints 3, Automated Reasoning 4, Probabilistic Reasoning 4, Machine Learning 5, Ethics 1, Exams 3, Std presentations 4 Activities CommentsJan 14 TAI an Introduction, and get-to-know quiz My thoughts on Grad projects for now  HYPERLINK "http://www.cs.fit.edu/~dmitra/ArtInt/" http://www.cs.fit.edu/~dmitra/ArtInt/Fall2019/GradProjectLists.docJan 16 RRequired CS background list Big O-notation (Algo-Intro slide 1-10), Djikstra algorithm (Algo-Graph slide 23-25), Min-spanning tree algorithm (Algo-Graph slide 54, 57-8) AI SEARCH (From my slides): 8-puzzle: https://n-puzzle-solver.appspot.com/ BFS- Uniform Cost; DFS-Depth Limited --MLK, Jr. memorial holiday Jan20Jan 21 TAI SEARCH (From my slides): 8-puzzle: https://n-puzzle-solver.appspot.com/ BFS- Uniform Cost; DFS-Depth Limited (up to sl 22) Iterative-deepening NP-completeness (Algo-Complexity slides 6-9:NP; 10-14: Problem class; 15-22: PvsNP; 32-33: CLK-thm; 41-46: NP-complete; 63-67: What-to-do-with-NPcomplete) Quiz-1 due Coding exc-1 announced. Due: Feb 11 in class Jan 23 RIn-class quiz on BFS, DFS Heuristic Search A*, IDA* MySlides 15-35 Text Slide Ch4aGraduate projects Described below this table Jan 28 TGame (Adversarial) search MySlides: 1-8 Local search from Text-slides Ch4b: Local search, Hill-climbing, Simulated Annealing, Local beam search, Genetic Algorithm, Gradient search, Remind imp of heuristics in PRUNING search tree; Jan 30 RMy SearchSlides continued (36-48) Take home quiz: Yes, if Bucharest->Timisoara A*-shortest path is not feasible to compute with given data, then compute the reverse, Timisoara to Bucharest. Search - exercise problems? REASONING WITH CONSTRAINTS: Motivating with Map/Graph coloring, Backtracking, Forward Checking (TextSlides)Grad students stay back a few minutes AGAIN after lecture today, I would like to know what have you gathered so farFeb 4 TCh5 text slidesTake home quiz dueFeb 6 RQuiz on search (15 min) Constraint Reasoning SPATIO-TEMPORAL CONSTRAINT REASONING from my slides. A relevant web page:  HYPERLINK "https://www.ics.uci.edu/~alspaugh/cls/shr/allen.html" https://www.ics.uci.edu/~alspaugh/cls/shr/allen.html Grad Project written proposal due in class, hard copy 2-3 pages (any format is ok) Feb 11 TNote: Input graph vs search tree vs control-flow-on-search-tree vs (shortest) path MySlides,; AC-3, SAT, and SAT-DPLL SAT in Algo-complexity slides: 34-8 Brief intro to artificial neural network: My slides 25, 28, 30, 32, 36-7, 44 AUTOMATED REASONING (ch 7): Basics & SyntaxCoding assignment-1 due Grad Project written proposal feedback My talk on Maze to Stanford on R-4pm on skypeFeb 13 RAUTOMATED REASONING (ch 7): Syntax-Semantics-Model, Satisfaction-Entailment-Inference procedure-Validity (up to TextSlide 41, ch-7) Assigned a take-homeFeb 18 TGrad Project status presentation (1/2 hr) Take-home quiz due More quiz assigned Grad Project status presentation, each group 5-7 mi, make sure that your media works in the classFeb 20 RAdversarial search: discussion AUTOMATED REASONING continued: Propositional Knowledge Base, Model checking algorithm, Forward chaining algo, Backward chaining algo (sl#40-66) CNF, Resolution Algo (p255), Horn Clause, Definite clause A sentence not in Horn form: P is not true, ~P=>TFeb 25 TPilot online test on Canvas over Search (5 min) AUTOMATED REASONING: Revisited PropLogic algorithms First Order Logic-Motivation;Feb 27 RAUTOMATED REASONING (possible lecture on Skype): Model-Interpretation-Quantifiers-Inferencing Completeness-Herbrand Universe, Mar 3 TQuiz on Search and Constraint reasoning (with online component) 15 min Revise, Frame problem-etc., (sl 17-28) Grad Project First Progress Report due: updated proposal (quite possible that your work has changed from your proposal mention that), including work that has been done so far and your plan for futureMarch 5, ROccur check: Substitution like x/S(x) not allowed in most logic programming languages, Unification-algo syntactically checks for that and fails in that case. Occur check makes it unsouond! Otherwise, x/S(S(x)) infinite looping source of semi-decidability of FOL Unification, Forward Chaining, Backward Chaining, Resolution strategies Homework: Use slides 25-27 of Inferencing with FOL. Show the steps of the proof for query Criminal(West). The slides animates the proof tree for your understanding. You write the steps explicitly as a Forward Chaining algorithm will work. Due in hard copy, next class. Grad Project status discussion (cancelled presentation) & reports (ungraded) returned! Undergraduate student attendance required (unless pre-excused)Undergrad Coding Exc-2 due April 6, Monday, 11:59pm Description is below this table Form 2 persons groups, except one group has to be singleMar 9-20-- Spring break --Grad Project status presentation (15 min each group, to be scheduled online, possibly March 26) Revised report (soft copy, via the Canvas) due, including sample output from your implementations, on the same day! Let us extend this through Wednesday 5pm EDT.Mar 19, RLectures will be most likely on Zoom during the class hours; link will be posted on Canvas; zoom is now linked from Canvas; maybe limited to 40 min per session; in that case we will go with multiple sessions; I will do a trial run at 5pm this day 3-19-R for a few minutes. Join the class (optional) if you can. Mar 24 TOnline lecture: will post zoom link on Canvas announcement just before the class. You must join with computer audio, and then mute yourself. There is a hand-raising feature on zoom. Also, you should have the chat open and use it if necessary. For asking question, you may unmute yourself. I will share my screen, and would like 2-3 folks to be on video at a time for a better feel. Others should turn off video to reduce bandwidth. REASONING WITH UNCERTAINTY: Ch 13: Motivation Probability-Joint-Inference-Conditional-Bayes rule; Zoom session continues: Three Graduate Project status presentations (8 min each group, one of the group member will take over the screen sharing on zoom, practice beforehand so that you may express succinctly your results within time). Organ Recognition, Indus ValleyHomework (on slides 25-27 of Inferencing with FOL) The Forward-Chaining algo is explained with animation on slides. You need to write down the deductive proof step by step. due: upload photocopy on the Canvas. This is also a new trial! Hopefully, all of you will submit. I hope Undergrad students paired up for the Coding Assignment-2. A discussion forum is created on Canvas for this respond there. I will check on the pair-names in class today.Mar 26 RTest on Automated-Reasoning/Logic online (15 min. after you join starting at 8+ pm, 10 questions) (Lecture for 20 min) Probabilistic reasoning contd. Ch 13: Sl 10-26 Two Graduate Project status presentations (8 min each, practice beforehand to express succinctly your results within time). Time-series topology, Maze-path decision problemProvide me feedback on the online delivery mechanism. I will take attendance next class onwards, except for those who are excused beforehand. The class is recorded and linked from the class page. Feel free to use Canvas mail to communicate with me, I may be on Canvas Chat MT 1:30-3:30p for OH. Mar 31 T Probabilistic reasoning contd. In-class quiz (Sample exercise Ch13: 13.8, 14, 15 ) Bayesian network Ch 14 starts Apr 2 RBayesian network Ch 14 contd. Possible in-class quiz. If you do not see my chat box open remind me.Undergraduate coding assignment (below) modified Graduate intermediate Presentation-Slides or Report due by 5 pm April 2Apr 7 TUndergrad Coding-2 status discussion MACHINE LEARNING: Decision tree, basics Apr 9 ROnline test on Probabilistic reasoning (20 minutes, 10 MCQ questions, Ch 13, 14). You may need a blank paper handy for rough work. MACHINE LEARNING: Decision tree, contd. Class will be recorded and shared as no one objected. Please remind me to turn on recording (Try to join with video on, zoom bandwidth seems not a problemApr 14 TCh 18.6: MACHINE LEARNING continued: MDL, PAC, 18.6.1-2. Regression: Linear, Multi-variate regression, linear In-class quiz Graduate Project status presentation: Motion-type identification Take Home on Probabilistic Reasoning (to publish during class-time today), due 4/16/R/7:45pmClarifications on Coding-2 (UG) being added below. I am planning to offer a Summer course Computational virology: A machine learning perspective to address some of the genomics related questions with viruses. Basics of genomics will be addressed.Apr 16 R18.6.3-4. Classifiers, Logistic regression, Perceptron 18.7.1-4. Artificial-neural Networks [Start from MySlide#24, exclude slides 39-43]  UG Coding Exc-2 due Monday-4/20-11:59pm (not 4/16/R)Apr 21 TArtificial-neural Networks [Start from MySlide#24, exclude slides 39-43] 18.7.1-4. Artificial-neural Networks Ch 18.8: non-parametric, kNN and variations Apr 23 RCh 18.8: non-parametric: LSH (start - slide 51), kernel-regression and 18.9-10 SVM DT tutorial continues Apr 28 T Clustering basics: K-means, from  HYPERLINK "https://en.wikipedia.org/wiki/K-means_clustering" https://en.wikipedia.org/wiki/K-means_clustering K-median, Slide 64 hierarchical, density-based Canvas problem: could not create Machine Learning (ML) exam, postponed for another class. A take-home for today, due 4/30/R/8pm Graduate groups demo starts: Chris on Time series clustering by persistence homology Graduate Project Final report due Monday 27th by 9pm on Canvas by e-mail (i) A report between 4-10 pages, including an abstract and references in NIPS conference format (https://neurips.cc/Conferences/2019/PaperInformation/StyleFiles), you need not use LaTex, but just follow the formatting style., (ii) Your source code including instructions on how to run and any dependencies and library details, and (iii) data files with metadata information (to understand data). All three in a zip file, named using: Sp20_project names signature id_one group members name Apr 30 RPlease do the instructor review. Test on ML, outside class hours: Available 4/30/R/9am Due 4/30/R/8pm just before the class --Download the Table-1 before starting the test https://cs.fit.edu/~dmitra/ArtInt/Spring2020/Table-1-ML-Test.docx Note: A smartphone may be too small a device to view everything for this test. An additional Table-1 will be provided as a supplement to answer some of the questions. (If I forget please e-mail me.) AI and Ethics (my slides are included in syllabus/tests) Exam: Predicate logic, Probabilistic reasoning, Machine Learning, Ethics Graduate groups demo continues DISCLAIMER: CANVAS HAS NO FORMULA NOW. I USE MY OWN SPREADSHEET. YOU WILL SEE YOUR COMPREHENSIVE GRADE THERE IN A SEPARATE COLUMN IN CANVAS AND LETTER GRADE IN PAWS.5/8/2020/RWe will continue with Grad demos during the exam period on Thursday 8:30pm-. Mandatory for undergraduates to attend. The exam will be for 30 minutes from your start time. Only online test, no take-home. You may take the test any time between 9am-8pm on 5/8/R (like the MLtest was). You will need 3 Tables from the class page, I suggest you download them before starting the exam. Having additional rough papers to work on may help. Feel free to use calculator but that may not be needed. Unfortunately, test answers will not be available soon, maybe next week! Good luck! Peace, health, and productivity! Exam period: Thursday, May 7, 8:30-10:30 p.m.All Graduate project reports look very good! However, I have some minor but strong suggestions. They are below in a table. I will expect them taken care of by the coming weekend. Grading weights at the top of the table. https://www.fit.edu/policies/final-examination-schedules/examination-schedules/spring-final-examination-schedule/Look up for evening classes, TR, in that web site8 p.m. ClassTuesday and ThursdayExam: Thursday, May 7, 8:30-10:30 p.m. GRADUATE PROJECTS (Project partners may be individually graded if necessary) SuggestionDone?Update reportMotion detection: Sample images of input as Fig. Graphic description of the network architecture.Extra data Possibly a 10 min presentationToMato: IEEE and/or Stanford group presentation? Segmentation output masks will be needed corresponding to each input image, for my archival & the paper.Update reportTime-series: More needed on your slice-by-slice algorithm: formal algorithm as a figure, and explanation with a running example. Add info on computation speed. Add country-to-country comparison as an appendix. Stanford group presentation? ArXiv submission?Update reportIVC: Provide numbers on used real data, then augmented data. Exact transformation math and the augmentation process used: rotation steps, translation steps and range, etc. I presume as you transformed data you also transformed the bounding box by the same transformation. Explain it.Update reportMaze: More on evaluation process needed: test set numbers by each case? Four sample-mazes as figures: (with & without path) X (successful, not) Table-2: compile or training time? will you update last paper and submit to any AI journal? Maze-decision problem: Our work so far was on the existence of path between coordinates (0,0) and (n-1,n-1) corners in an n x n maze. We would like to do the same between (0,0) and (x,y) for variable x,y Middle layers -> eight numbers. What I see happening is a disconnect. You apparently do have a machine learning model that somehow detects the text box or boxes. I do not understand how? One way would be to draw your neural network's architecture and explain what is going on. If you do really have a solution to the problem of "text box recognition" that is great! I, and you, need to understand what is going on. I asked you to submit a report clarifying this. Deep learning of activities-recognition: Any available activities database available online with labels that you would like to work with? (Yes, I found Microsoft has a kinnect image database; NIST-IARPA-challenge (ActEV) database? Any other?) How will you augment data for training? Which computer will you work on? May need large memory, fast processor and GPU? Mathematical model or category-classification? I prefer category classification (car entering/existing parking lot, backhand/forehand stroke, etc.) Tim Porath (and Sarah Arend?) After 1st presentation Did you converge on a data set and validation plan? Training code? Input layer format matching with your data? Clustering time-series based on topology: Persistence-based clustering of time-series. Cancer image from contrast-enhanced ultrasound (CEUS) data. (No knowledge of cancer or ultrasound is needed) (1) Learn about time-series. There are plenty of time-series databases, the most known one is the Univ CA-Riverside (UCR) database.  HYPERLINK "https://www.cs.ucr.edu/%7Eeamonn/time_series_data_2018/" https://www.cs.ucr.edu/%7Eeamonn/time_series_data_2018/ Even though, you will possibly work with our medical data. (2) How to compare two time-series, or what are the types of distance between two time-series? (3) Learn Topological data analysis (TDA) and persistence homology (PH). (4) Apply PH to cluster time-series based on their shapes. (5) If possible, compare against another clustering technique, e.g., K-means clustering. Chris Woodle (and Valerie Kobzarenko) After 1st presentation Is the code for time-series to PD ready? Debug with hand generated cases. PD to PD comparison (after removing diagonal-hugging points) is point-cloud comparison. Which software to use? Algorithm? Sum-of-min? Topology based clustering of organ voxels in a whole-body CT image (2D or 3D?): Learn TDA and ToMato software, use it on large whole-body image. Try to automatically recognize organs, based on persistence homology (PH, encodes shape). (1) Download ImageJ first, to visualize data. (2) Learn PH as it represents shape of an object. (3) Learn density-based clustering, ToMato software uses that. (4) ToMato has two implementations available on the Internet: Python-based, works only on 2D (can we modify source to make it work on 3D?); and C++ that works on nD but very difficult to install for library mismatch (not maintained). Check which one you would like to work with. Two students last semester made the C++ version work, we can seek help from them if needed. (5) Try to find multi-organ 3D medical image online, any animal or any medical imaging modality will do. For that matter, any image with isolated (should have complicated shape) clusters will do, as long as you can map it to a shape-based-object-recognition problem. After some success from you, we will release human anatomy data to you. Mircia Silaghi and Ran Bi After 1st presentation Did you obtain organ/ROI segmentation? Wise tool is another candidate. PD for whole-body image? Any pattern for each organ look over PDs for multiple images. Template PD point cloud should let you recognize organs. Needs PD -> image mapping. After 2nd presentation & Intermediate report Figures should have self-content captions with as much explanations as possible. Semantically organize them under same figure number, Fig 1a, b, ... Refer the Fig number within the text. I am not very clear on your Figure p8 if it removes kidney. Looks more like merger of two organs, upper kidney + liver or spleen Is PD translation invariant? Rotation and scale? Future idea (Not in your project): Can we have a GUI that would visualize the clustering on left panel and the PD used on the right. Moreover, the parameters can be changed on sliding scales, d value, the diagonal for cutoff, others. One mouse overs a PD point and it highlights the cluster on left, and one mouse-overs on a cluster and it shows the PD on the left. Discovering approximate symmetry group, and symmetry axes (3D protein) of objects with deep learning: CODING EXERCISES 1. All students, individual assignment: Implement or use library for three data structures: queue, stack, and priority queue with integers. Note, queue and stack ignores any number or weight in graph. Use these data structures to develop three graph-search algorithms over input: a graph, a source node, and a goal node. Three data structures from the above should be pluggable in your code. There will be three search algorithms with three data structures. Use your search algorithms to input the Romanian road-network map from the textbook (Fig. 3.2, p68), the start node as Bucharest and the goal node as Timisoara. Children from the root Bucharest to be expanded are in this order: Urziceni, Fagaras, Giurgiu, and then Pitesti. Note, the priority queue may rearrange the node orders, which is ok. The priority queue will use the shortest distances from Bucharest to the nodes in it using the road-distances on the map (NOT the straight-line-distances) and hence, should find a shortest path. Your algorithm should print the nodes as they are traversed by your algorithm. Submission in hard copy: a) Source codes; b) Full search tree by running each of your search algorithms. Due: Feb 11 in class [20 pts] 2. Only Undergraduate Students assignment 2 (in group of two students): Code the min-max algorithm for the adversarial game search with and without alpha-beta pruning, Input: NxN tic-tac-toe with varying N between 0 through as much as you can handle. For N>5, implement fixed ply (depth of search) at 5 3. You will need a board evaluation function at the leaf node of the search tree. Let the heuristic value be normalized to [-1, 1]. Output: 1) Play yourself (human-computer) against it print sample game for N=7. [All real play outputs are to be with alpha-beta pruning.] 2) Play against some other group (recorded jointly with the other group for the same game), with N=10. 2) Let your code play against itself (two copies of your code, computer-computer), for N=9. Same output. 3) Record CPU times for play-against-itself N=3, 5, 9 Submission on the Canvas: source code, print-outs, and a report. [30 pts] Due: April 16-Thursday-11:59pm. ASK FOR ANY CLARIFICATION. 1) Many are finding that the complexity is too high for N>3. In that case use ply=3, i.e., evaluate board after 3 recursive calls. You should be able to 3^2 depth that way (theoretically :) 2) There is a rule-of-game issue. Say for N=4, there are 4 ways to win/loose on horizontal rows, similar, for columns. However, there should be only 2 ways to win/loose with diagonals, for any N. 3) Dont forget to implement the alpha-beta pruning. That saves a lot of computational time! I modified the assignment above to exclude without alpha-beta pruning part. UG Groups: Emran - Siyu Fu Erika - Rishob Thanart Stian Hunter Jose Nic Taylor Andrew Alternative: Research on quantum neural nets.  $%'-0234>U[]cdgjprs򽯢rj`Z h CJjh CJUh0JCJjhCJUjhCJU he#CJ heCJ hCJ h>*CJhcYhCJmH $sH $hcYh>*CJmH $sH $h &>*CJmH $sH $ h81U5\ hLF5\ hR5\ h$ 5\ h]5\ hm%5\ h5\ hm45\!34U$ P _ C I X Y f $Ifgd<K$$IfK$gde#$a$gd;$a$" # $ 0 2 4 5 8 < O P \ ^ _ n r s u w ýývlblXNhher5CJhh*CJ hAMCJh.h 0JCJjh CJU h CJh h CJ    ! & > B C Y f g i j s w z ù͛ץ}q}gaXRX h ixCJh!h!CJ h]CJh|h]5CJhh<57CJh|h<5CJhhm5CJhh<5CJhh%.5CJhh5CJhh+n5CJhh5J5CJhh55CJhhLC5CJhhC5CJhh`5CJhhR5CJhh+d5CJf g h i j |t2--gd!AkdU$$If!B t634Bayt]$Ifkd$IfK$L$K\ceBBBB t634Bayt<z   4 5 @ A v w   ( - =mnoýôϟϙϓϟϓϓύ~ung` h%56 h856 h!56he#hqRCJhqRhqRCJ hqRCJ hGOCJ h%fCJ h83CJ hX$_CJh!hFCJ h!CJhFhFCJ hFCJ h,CJ hFpCJh!h!CJ hhlCJ h{CJ h,5CJh?!h!5CJh?!hds$%0N츴ۋzof]ho!5CJaJh?!5CJaJhv`h?!CJaJh'g6CJaJh'gCJaJh?!h?!CJaJhD}dhD}d5CJaJhD}d6CJaJhD}dCJaJh?!hv?h?!6CJaJh h?!6CJaJhMh?!CJaJh?!CJaJhu+6CJaJh?!6CJaJhwE h?!CJaJ"pggg $Ifgd&kd$$IflF8r0::$ t0:    44 laytS7'?s$%0pggg^ggggg $Ifgd'g $Ifgd&kds$$IflF8r0::$ t0:    44 laytS7' 01^_`ig^^^^^ $Ifgd,kd$$IflF8r0::$ t0:    44 laytS7' $Ifgd& N]^_`hiǾwi`TI>h?!56CJaJh?w56CJaJh'gh'x6CJaJh'x6CJaJhv?h?!56CJaJh'x56CJaJh1h?!CJaJhmlhO;CJaJhv?h?!6CJaJh'g6CJaJh?!CJaJh?!6CJaJhO;6CJaJhJ;h?!CJaJhJ;CJaJh?!h{h?!6CJaJh ho!5CJaJh+N5CJaJ ^ULC $IfgdC $IfgdO; $Ifgd}kd$$IflF8r0::$ t0:    44 laytS7' $Ifgd(0 $IfgdB!=>C   ǻskc[[SH[hmlhO;CJaJhFCJaJhO;CJaJh\cCJaJh?!CJaJhv?h?!56CJaJh4!h'g6CJaJh!6CJaJhCCJaJh6=CJaJhh6=6CJaJh6=6CJaJh'gh6=6CJaJhJ;h?!CJaJhJ;CJaJh?!hGh?!5CJaJh?!56CJaJh'g56CJaJULCC $IfgdO; $Ifgd<)kd,$$IflF8r0::$ t0:    44 laytS7' $Ifgd4! $Ifgd\c $Ifgd!45~rgYUMD8hN)'h?!6CJaJh:\6CJaJhJ;CJaJh?!hh[n56CJaJhC56CJaJhChC6CJaJhC6CJaJh,=6CJaJh,=hC6CJaJhmlhC6CJaJh"CJaJh?!CJaJh'rCJaJhCCJaJhL/3CJaJhCJaJh"56CJaJ#hUhU56B*CJaJphhU56CJaJ5a^U $IfgdQkd$$IflF8r0::$ t0:    44 laytS7' $IfgdC $IfgdO; E[RI $Ifgd8(j $IfgdQkdR$$IflF8r0::$ t0:    44 laytS7' $$Ifa$gdQ $Ifgd%CDZ[ĶppbpTH<htth?!6CJaJhmlh?!6CJaJh%h?!67CJaJh%h8(j0J7CJaJ jh%h8(j7CJUaJh%h8(j57CJaJh%h8(j7CJaJh#h#CJaJh#CJaJh%CJaJh%h%56CJaJh%h%5CJaJhJ;h?!CJaJhJ;CJaJh?!hoNehU56CJaJh56CJaJEZ*+,-Fkd$$IflF8r0::$ t0:    44 laytS7' $$Ifa$gdv $$Ifa$gdQ $IfgdE=H $Ifgd8(j+,-56  Źuiu]Q]Q]QEh)hW&m5CJaJh)hS55CJaJh)h#5CJaJh)h25CJaJh)hy5CJaJh)h#CJaJh)h2CJaJh)hnCJaJh)h~ ~CJaJh)h#6CJaJh)h~ ~6CJaJhJ;h?!CJaJhJ;CJaJh?!hv?h?!CJaJhoNeh?!56CJaJh?!56CJaJ-6 !Ngh $Ifgd4! $IfgdJ) $Ifgd*d $Ifgd# $IfgdQ  !>GMNefhmu}0DKLǽxpeZODhS|h?!CJaJhS|hYCJaJhS|h$CJaJhS|hx ,CJaJh?!CJaJhJ;CJaJh?!hv?h6CJaJh656CJaJhoNeh?!56CJaJh?!56CJaJh\ h?!56h\ hv56h=h?!CJaJh)hJ)CJaJh)h!CJaJh)hx ,CJaJh)h#5CJaJLabpggg[ $$Ifa$gd4! $Ifgd4!kdx$$IflF8r0::$ t0:    44 laytS7'L`abckl!"#+,ŻznbzbTIAh#$CJaJh"h#$CJaJh7hb56CJaJh7h -6CJaJh7hb6CJaJh7h?!6CJaJh7hHF6CJaJh7hc#6CJaJh}h?!CJaJh t h th t hN56h th t56hJ;h?!CJaJhJ;hJ;CJaJh?!h1.h?!6CJaJhc#hc#5hS|hc#5CJaJbclpg^^ULLL $Ifgdb $Ifgd$ $Ifgd*d $Ifgdpx6kd $$IflF8r0::$ t0:    44 laytS7'"#,KLo^ULLL $Ifgd#$ $Ifgdpx6kd$$IflF8r0::$ t0:    44 laytS7' $Ifgd!> $Ifgdbo;QR[Ukd1 $$IflF8r0::$ t0:    44 laytS7' $IfgdX $Ifgd*d $Ifgd#$ :;PQRZ[klmskcWKhK#h?!5CJaJhK#h~5CJaJh"CJaJh CJaJhDh?!56CJaJhSh?!CJaJhm(CJaJh{CJaJh]CJaJh]h]5CJaJh{5CJaJh]5CJaJhJ;CJaJh?!hnh?!6CJaJhh 6CJaJh?!6CJaJhh?!5CJaJh?!CJaJ[KUkd $$IflF8r0::$ t0:    44 laytS7' $Ifgdi $Ifgd" $Ifgd*d KlmnowULCC $Ifgd7 $IfgdikdW $$IflF8r0::$ t0:    44 laytS7' $Ifgd" $Ifgd $Ifgd~mnovwr'F^}yqe]U]M]Ehm@CJaJh%CJaJh7o(CJaJhCJaJhM1h6CJaJhEjCJaJh> ih?!56CJaJh 56CJaJh"56CJaJh~h?!CJaJhVCJaJh5?CJaJh~CJaJh_v\CJaJhK#CJaJhK#5CJaJhJ;h?!CJaJhJ;CJaJ hgnBh?!h}h?!5CJaJ9sLkd $$IflF8r0::$ t0:    44 laytS7' $Ifgd' $Ifgd" $Ifgd~ $Ifgd7~  "!#!z!!! "G" $Ifgd" $Ifgd7 $Ifgd $Ifgdi }~  !!!"!#!7!Y!Z!y!z!!!!!!!!!! "÷{oco\QFQhEj56CJaJh&56CJaJ hEjh:PhHh76CJaJhHhhU6CJaJhHh:P6CJaJhHh,6CJaJhHh&6CJaJhHh.m6CJaJhHh6CJaJhHhEj6CJaJ h56hh6CJaJh~6CJaJh6CJaJhCJaJhCJaJhCJaJ "G"H"P"T"Z"d"e"f"g"""""""""""#:#;#i#j#k#l#ž}}o}a}o}VLžhZARhL56hXh56CJaJhqdhZAR56CJaJhqdh\56CJaJhqdhf56CJaJhqdh>b567CJaJhqdh>b56CJaJhqdh56CJaJh?!CJaJ hgnBh?!h?!56CJaJhJ;6CJaJhdoh?!6CJaJh>b6CJaJ hgnBhEjhQm56CJaJG"H"Q"d"e"pg^g $IfgdJ; $Ifgd4!kd} $$IflF8r0::$ t0:    44 laytS7'e"f"g""j#k#pg^UL $Ifgd4! $IfgdL $Ifgd $IfgdLkd $$IflF8r0::$ t0:    44 laytS7'k#l#v#H$$$$pg^^UL $Ifgd7 $Ifgd $IfgdW $IfgdLkd $$IflF8r0::$ t0:    44 laytS7'l#u#v#H$a$t$$$$$$$$ % %G%H%I%Y%`%x%%%%%&ѴynbnyWLWAWAWnhBR!hxMCJaJhBR!h6:CJaJhBR!h6CJaJhBR!h p>*CJaJhBR!h pCJaJhBR!h0XCJaJh?!CJaJhJ;CJaJ hgnBhWh756CJaJhqdhW56CJaJh h*X56CJaJh hW56>*CJaJh hW56CJaJh hWCJaJhWhW6CJaJhW6CJaJ$$$I%%l&m&&&&pg^^^^^UU $Ifgd $Ifgd>b $IfgdXkd6 $$IflF8r0::$ t0:    44 laytS7' &&&&8&l&m&o&&&&&&&!'''{''''''''''Ͷͫ|ppbVKh>b56CJaJhqdhqd6CJaJh.<hAU56CJaJh.<h5CJaJh.<hJl5CJaJh.<h65CJaJh.<hqd5CJaJh.<h0XCJaJh.<hqdCJaJh.<h>b6CJaJh.<hCJaJh.<h>bCJaJ h hxMhBR!hIu{CJaJhBR!h pCJaJhBR!h_RCJaJ&&'' (2(V((((((()5)])x)) $IfgdXh $Ifgd $Ifgd' $IfgdAU $Ifgdqd $Ifgd'' ( (((U(V((((((((4)w))))))))))·¬ugYKYhS7'hJEM56CJaJhS7'h>b56CJaJhS7'h 56CJaJhB,h?!CJaJhJ;CJaJh~h?!mH $sH $hh56CJaJh;56CJaJhXh56CJaJh756CJaJh56CJaJh>b56CJaJh0X56CJaJh0X6CJaJh6CJaJh56CJaJ)))**)*H*X*Y*pggg^^^U $Ifgd7C $Ifgd,Z $IfgdXkd $$IflF8r0::$ t0:    44 laytS7'))****)*6*O*X*Y*\*******++;+~r~r~dVH?hb 6CJaJhqdhqd56CJaJh(h(56CJaJhS7'hAU56CJaJhS7'hJl5CJaJhS7'hqd5CJaJhS7'hS7'5CJaJhS7'hqdCJaJhS7'h,Z6CJaJhS7'h,ZCJaJh UCJaJhS7'hJEMCJaJhS7'h>bCJaJhS7'h>b56CJaJhS7'hJEM56CJaJhS7'hqd56CJaJY**+<+g+++++,0,1, $Ifgd>b $Ifgd( $Ifgdqd ;+<++++0,1,2,:,;,<,Z,[,\,],k,l,,,,,,,,,,,Ž|qe]OEhih?!56h8Jh?!56CJaJhHhH5hfhH5CJaJhfh?!CJaJhfh U6CJaJhfhH6CJaJhBR!CJaJhfhHCJaJhfh UCJaJhz[h CJaJh?!CJaJhJ;CJaJh~h?!mH $sH $hn h 6CJaJh 6CJaJhb 6CJaJh?!6CJaJ1,2,;,<,\,j,k,,,,pgg^^^^^^ $Ifgd U $Ifgdz[kd\$$IflF8r0::$ t0:    44 laytS7' ,,,,,,, -A-g^UUUUL $Ifgd4! $Ifgdw $Ifgd0kd$$IflF8r0::$ t0:    44 laytS7' $IfgdD,,,,,,,,- -S-i-j-v-----------Ĺzrj^RDh\h$56CJaJh\h?!5CJaJh\hH5CJaJh?!CJaJhJ;CJaJhUh?!CJaJhUCJaJhChg`5CJaJhChU5CJaJhChu5CJaJhfhBR!5hHhBR!CJaJhHh?!CJaJhHhH5CJaJhHhHCJaJhJ;h?!CJaJhJ;hJ;CJaJh?!A-R-S-j------g^UU $IfgdH $Ifgd0kd$$IflF8r0::$ t0:    44 laytS7' $Ifgd4!------...ULLL $Ifgd Ukd$$IflF8r0::$ t0:    44 laytS7' $Ifgd4! $Ifgd\ $IfgdH-------...... / / /$/I/J/K/S/T/y/ǻ{qg`TI>h0ah UCJaJh<h?!CJaJhIh?!6CJaJ hfM56hihD56h\h?!56 jJhH5h\56CJaJhH5h\56CJaJhH5hHCJaJhH5h UCJaJhH5h\CJaJhH5h\5CJaJh?!CJaJhJ;CJaJh?!h8Jh?!56CJaJhHhH5CJaJh\hHCJaJ... /'/J/K/T//UL $Ifgdqkd$$IflF8r0::$ t0:    44 laytS7' $Ifgd(= $Ifgd\ $Ifgd Uy/|//////00o0p00011j1k1l1t1u122ƺxpkp_[SK@9 h0ah?!h^Ch UCJaJh?!CJaJhJ;CJaJh?!hx h%6CJaJ h_ 6h_ h%6h_ h?!6h_ hH56h<^h<^CJaJh0ah<^CJaJh0ah<^56CJaJh0ah56CJaJh0ah5CJaJh0ahCJaJh0ah<^6CJaJh0ah?!CJaJh0ah UCJaJh0ah U7CJaJ/////00p00001<1k1 $Ifgd& $Ifgd U k1l1u1222,292pg^^ULC $Ifgd4! $IfgdS7' $Ifgd& $Ifgd U $Ifgd*d kd;$$IflF8r0::$ t0:    44 laytS7'2222#2+2,28292:2B2C222222222?3ǻtf^VK@h{W hCJaJhJ;h?!CJaJhS7'CJaJh`h?!5h`h?!56CJaJhyh?!CJaJhCJaJh^ChCJaJh?!CJaJhJ;CJaJh?!h`h?!6CJaJh)/]6CJaJh`hS7'6CJaJh U5CJaJh)/]h)/]5>*CJaJh)/]5CJaJh?!56CJaJh& h?!56CJaJ92:2C222222pg^UUUL $IfgdbQ $Ifgd $Ifgd U $Ifgd4!kd$$IflF8r0::$ t0:    44 laytS7'222?3@3A3W3X3Y3Z3pg^^g^UUU $IfgdS7' $Ifgd $Ifgd*d kda$$IflF8r0::$ t0:    44 laytS7' ?3@3A3B3L3U3V3W3X3Y3Z3[3c3d3e33333333(4)4˽}rcrcVHcr=hhvCJaJhhK00J>*CJaJhhK00JCJaJjhhK0CJUaJhhK0CJaJh!=hj7CJaJh?!CJaJhS7'CJaJh?!hv?h?!5CJaJhS7'56CJaJhS7'6CJaJhN!h?!56CJaJh{W h56CJaJh{W hv56CJaJh{W hbQ56CJaJh{W h^CCJaJZ3[3d3e33)4*4444pg^U^^ggg $IfgdK0 $Ifgd U $Ifgd;kd$$IflF8r0::$ t0:    44 laytS7' )4*444444445555%5.50595B5C5M5n5o555555ΤyiyYyNyCyCyCyh`56B* phPh.56B* phPh.hbQ567B* phPh|YhbQ56B* H*phPh|YhbQ56B* phPh|YhbQ56h>_!h?!7CJaJ hYh?!hh56CJaJhh56CJaJhh^C56CJaJh56CJaJhh?!56CJaJhh{W 56CJaJhhK0CJaJ455595o55536T666667?7@7 $IfgdbQ $Ifgd4! $Ifgd; $Ifgd55162636S6T6777 777>7?7@7A7J7T7X7k7l77777777777ٶzpibXiXiXiQ h{956hh56 h{W 56 h56hh{W 56 hE56h`56>*h`hE56>*hS7'CJaJh?!hv?h?!CJaJhbQ56CJaJh^C56CJaJh|Yh^C56h`hbQ56B* phPh|YhbQ56h|YhbQ56B* phPh`h`56B* phP@7A7J7k7l77777:8pg^^^^^^^ $Ifgd{W $Ifgd;kd$$IflF8r0::$ t0:    44 laytS7' 798:888889999'9:9;9<99999L:M:N:X:Y:k:z:::::::ºxndZdPdnPhphp56hph[v056hph}T56hphE56hjhjCJaJh0sCJaJhS7'hv?hS7'CJaJhS7'56CJaJ hhhh56hhh5hh Uhh U>* hh U hh{W hEhEha_hE h{956h{9h{956:8899;9<9999999M: $IfgdS7' $Ifgd; $Ifgd U $IfgdE M:N:Y:::t;<<<<pg^^^^^UU $IfgdS7' $Ifgda $Ifgd;kd$$IflF8r0::$ t0:    44 laytS7' :::::::::::;;%;(;-;K;O;V;\;];f;m;n;q;s;t;;;<<<<<<c=d=r======Żųгл쟗}v}v}v}o hj56 he56h hu'56hE56CJaJ hp5hph/ud5hphj56hph56hp56>*hph/ud56hph/ud56>*hphp56>* hp56hph[v056hphp56hpho)56*<=H=d========gkd$$IflF8r0::$ t0:    44 laytS7' $Ifgdi =======<>=>E>I>P>Z>\>n>o>p>>>>>>>>>·tl`RGh:h:CJaJh(0h:56CJaJh(0h(05CJaJh:CJaJhh/ud5CJaJhht5CJaJhth?!hv?h?!CJaJhtCJaJhCJaJh?!CJaJhh?!CJaJhthtCJaJh]Dh?!CJaJhjhu'6CJaJh hu'56h h56 h56===>o>p>}>>>^XXX$Ifkd@$$IflF8r0::$ t0:    44 laytS7' $Ifgdt $Ifgd4!>>>>? ? ???e`[SSGGG $$Ifa$gd $a$gd:gdr+*$a$kd$$IfF$FF Fv t0:t    34Bab ytt>>>>>>??? ? ???*???T@b@gAtABBCCCCCCCC D DDtDyDDDDʿʫ~s~sg~s~s~h?h3uF6CJaJh?h'wCJaJh?h3uFCJaJh?hO~CJaJh?h}rCJaJh?h$>*CJaJh$CJaJh hCT6CJaJh hCTCJaJhCTCJaJh:CJaJh4_CJaJh:h:CJaJh:h XCJaJh XCJaJ%??*???`TTT $$Ifa$gd kd$$IflFT,"f t06    44 lapyt ?????R@S@`TTTTT $$Ifa$gd kdI$$IflFT,"f t06    44 lapyt S@T@b@@@A6AeAfA`TTTTTTT $$Ifa$gd kd$$IflFT,"f t06    44 lapyt fAgAuAA"BBB`TTTTT $$Ifa$gd kd$$IflFT,"f t06    44 lapyt BBBB2CUCCC`TTTTTT $$Ifa$gd kd_$$IflFT,"f t06    44 lapyt CCCCVGGGGHKH`XXXXPPPP$a$gd$a$gd:kd$$IflFT,"f t06    44 lapyt  DDDEE5EEEEEEE>FVFaFkFlFmF5GUGGGGGGGGGGHHH꼱ymbSbh5h55>*CJH*aJh55>*CJaJh?h56CJaJhL6CJaJh?h_Q6CJaJh?h9lg6CJaJh?h$>*CJaJh5hMCJaJh?h?PCJaJh?h dxCJaJh?hTCJaJh?h6CJaJh?h3uF6CJaJh?h3uFCJaJh?hCJaJHHyHHHHHHHHII)I[IcIgIIIIIIIIII"JJJJJJJJ˿{{{ppph]h]hUhCJaJh?hDCJaJhDCJaJh?h"CJaJh?h|CJaJh?hu6CJaJh?h>6CJaJh?h>CJaJh?hCJaJh?huCJaJh?hu>*CJaJh?h5>*CJaJh5h9{LCJaJh9{LCJaJh5CJaJh5h_Q5>*CJaJ KHyHHHI#JKKKKKL'L(L?LLL+MMM N4N5NNNKOO$a$gd,Z$a$gdu$a$gdJJKK K KKK#K'K,KqKKKKKKKKKKKKKLLL&L(L>L徲skckXh5>*CJaJh,ZCJaJh5CJaJh5h55>*CJaJh5h55>*CJH*aJh55>*CJaJhu>*CJaJh?hu6>*CJaJh?hc6CJaJh?hU}6CJaJh\3CJaJh?h>CJaJhDCJaJh?h"CJaJhCJaJh. CJaJ>L?LNLULOOOOOOP PPPP0P1PBPrPPPPPPPPPȼxmbWbOGObWmh.qCJaJh6CJaJh?hC7CJaJh?hCJaJh?hrBCJaJh?h CJaJh?hr2CJaJh?h6CJaJh?hCJaJh?h dx>*CJaJh?h?P>*CJaJh?h>*CJaJh5h,ZCJaJh<h6CJaJh,ZCJaJh,Zh,ZCJaJhh5>*CJaJOOOOOPRQQRRRPRRRQSSVTTT>UyUUUUV$a$gd`]$a$gdhgdh$a$gd.$a$gd$a$gd,ZPPPPQQQQ!QRQQQQQRRR RRRR!R:RPRRRRS{me]e]RJFhhhhhh>*h?h.CJaJh9{LCJaJh.CJaJh5h.5>*CJaJh5h.5>*CJH*aJh.5>*CJaJhCJaJh?h8 ?6CJaJh?hS86CJaJh?hc6CJaJh?hCJaJh?h)CJaJh?h6CJaJh?hC7CJaJh?hrBCJaJh83CJaJSSSTTTTTUTVTUUVVVVVVVW$W.W/W0W1W8W*CJaJh?h>*CJaJh?hT>*CJaJh?hhCJaJh`]h5h`]5>*CJaJh5h`]5>*CJH*aJh`]5>*CJaJh9{Lhhh6h^90J h^9h^9h^9jh^9UVZVVVWW0XoXY5[O[P[g[[\[\\\\\ ]F]G]]]]]U^$a$gdpJ$a$gdv'i$a$gd:$a$gdhWWWWWWWWWWWWWWW XdXhXoXXXXgYhYYYY%Z9ZJZuZZZZZZZZ4[5[C[G[H[ɾ߾߾߾߾߾߾yh?h6CJaJh?h_Q6CJaJh?hTUCJaJh?hyCJaJh?hQ^U6CJaJh?hQ^UCJaJh?hrBCJaJh?hKnCJaJh?hmXCJaJh?hD:CJaJh?hCJaJh?h;PCJaJ*H[N[O[P[W[Y[f[g[[\\\r\\\]]]]]]l_m_y_____ҵ{tf[fPHh+Nh 5h:h:CJaJh>7>*CJaJhRh>7>*CJaJ hVhv'i hVhVhpJhV hVhpJhVh5>*hVhpJ5>*hpJCJaJhv'iCJaJh5hv'i5>*CJaJh5hv'i5>*CJH*aJhv'i5>*CJaJhvCJaJh?h_Q6>*CJaJh?h8 ?6CJaJU^^_l_m____dddXffMggg;h[hvh4iijjjj$a$gd3W$a$gd(0$a$$a$gd>$a$gd:$a$gdpJ______ ` ` ``*`@`A`a`````````aaa$a'a,akalaaaa'b1bFbObZbhbʾʲzoooooddzhh 4CJaJhh+NCJaJhhMKbCJaJhhQ@CJaJhh6d6CJaJhh CJaJhh6dCJaJhh6d5CJaJhh0E5CJaJhhj:5CJaJhhj:CJaJhh6LCJaJhh6VCJaJh+Nh6L5&hbqbrbbbbbbbc?cHc[cccccdd'd0d2dWdYdbdgdoddddddƺƯݍсѺѺvk]hh,G6>*CJaJhh6LCJaJhhj:CJaJhh06CJaJhh6d>*CJaJhhICJaJhhWCJaJhhO6CJaJhh 46CJaJhh 4CJaJhh6d6CJaJhh6dCJaJhhMKbCJaJhhMKb6CJaJddddddddddddde,e7eLeeeeeeeeeeeeee%f@fMf^fdfqffffffº}sn}}}g hhrr h6hhh67h&hh6h&hv'i6 hhhhh6d7 hh6d hhv'ihh#5hhwE75hh^05 hh hoMCJaJhh6dCJaJhhaeCJaJhhMCJaJhh+N6>*CJaJ(ffffMgNgqgsgggggggggggg hhh1h2h:h;h@hFhGhIhQhZhchghuhvhiAjOjTj`jdjnjtjjjj濸}uhv'iCJaJhh6 h{<6 hc6 h*76 h6 hV6 hr+6 hxa6hh6hhr+6 hh hh8 hhV hhwE7 hhr+h& hhhhfh&hv'i7 hhrr hhv'i.jjjjjjjjjkkk/k0k1k2k3k4k5k6k÷{h#hVOCJaJhVOCJaJhFCJaJh'CJaJh6VCJaJh^>*CJaJhQ@CJaJhv'ih_7CJaJhv'ih: 7CJaJ hV6 h3W6 hM6hMhM6h3Whv'i6h3Wh3W6h3Wh3W6>*jjjjjk/k0k1k2k3k5k6k$a$gdF$a$$a$gd3W ,1h/ =!"#$% DyK dmitra@zach. fit.eduyK 8mailto:dmitra@zach. fit.edu|$IfK$L$!vh#vc#v #vg#v2 :V K t6,534Byt<c$$If!vh#v":V  t6,534Byt]$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If!vh#v#v:#v$:V l t0:55:5$ytS7'$$If&!vh#v#v #vv:V  t0:t,55 5v/ 34Bytt$$If!vh#v#vw#va:V l t06555fpyt $$If!vh#v#vw#va:V l t06555fpyt $$If!vh#v#vw#va:V l t06555fpyt $$If!vh#v#vw#va:V l t06555fpyt $$If!vh#v#vw#va:V l t06555fpyt $$If!vh#v#vw#va:V l t06555fpyt s02 0@P`p2( 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p8XV~ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@_HmH nH sH tH @`@ NormalCJ_HaJmH sH tH DA D Default Paragraph FontVi@V  Table Normal :V 44 la (k (No List 6U`6 Hyperlink >*B*phHH 0o Balloon TextCJOJQJ^JaJN/N 0oBalloon Text CharCJOJQJ^JaJFV !F _FollowedHyperlink >*B*phOrj3j VO Table Grid7:V0B^@BB E0 Normal (Web)dd[$\$PK![Content_Types].xmlN0EH-J@%ǎǢ|ș$زULTB l,3;rØJB+$G]7O٭VvnB`2ǃ,!"E3p#9GQd; H xuv 0F[,F᚜K sO'3w #vfSVbsؠyX p5veuw 1z@ l,i!b I jZ2|9L$Z15xl.(zm${d:\@'23œln$^-@^i?D&|#td!6lġB"&63yy@t!HjpU*yeXry3~{s:FXI O5Y[Y!}S˪.7bd|n]671. tn/w/+[t6}PsںsL. J;̊iN $AI)t2 Lmx:(}\-i*xQCJuWl'QyI@ھ m2DBAR4 w¢naQ`ԲɁ W=0#xBdT/.3-F>bYL%׭˓KK 6HhfPQ=h)GBms]_Ԡ'CZѨys v@c])h7Jهic?FS.NP$ e&\Ӏ+I "'%QÕ@c![paAV.9Hd<ӮHVX*%A{Yr Aբ pxSL9":3U5U NC(p%u@;[d`4)]t#9M4W=P5*f̰lk<_X-C wT%Ժ}B% Y,] A̠&oʰŨ; \lc`|,bUvPK! ѐ'theme/theme/_rels/themeManager.xml.relsM 0wooӺ&݈Э5 6?$Q ,.aic21h:qm@RN;d`o7gK(M&$R(.1r'JЊT8V"AȻHu}|$b{P8g/]QAsم(#L[PK-![Content_Types].xmlPK-!֧6 0_rels/.relsPK-!kytheme/theme/themeManager.xmlPK-!R%theme/theme/theme1.xmlPK-! ѐ' theme/theme/_rels/themeManager.xml.relsPK] 6c> z NLm} "l#&');+,-y/2?3)457:=>DHJ>LPSWH[_hbdfj6k689;>?BFHJMOQSWZ]^bdfhjmpruxz|~f 0E-bo[KG"e"k#$&)Y*1,,A--./k1922Z34@7:8M:<=>??S@fABCKHOVU^j6k7:<=@ACDEGIKLNPRTUVXY[\_`acegiklnoqstvwy{}r""V|Z+++KLTL6cXXXXXX8@0(  B S  ?hlACLN 4 8 l n  "   )3688AX`$(t|) - H"J"O"Q"$$$$D&F&T'V'******N-O-. .$/*/g3m3555577S9X9H<K<Z<]<A A!A+A,A6AB BBBBCCCCCHHHHIIIJKKMMMM@OFOTPZPsPyPQQ5S;S+?+@+V+W+W+X+X+Y+Y+`+a+b+c+d+d+e+e+++(,),*,,,-----,-.-9-B-C-M-n-o-y-y-------///// /7/>/?/?/A/1:1111L2X2Y2445p6p6667;%D&D'DGKULZTkW\\\\b]]]__2`:`bbc5c8c#PPw!BggRlmmww"rHNPffgijjl!!##0$:$;$<$$$$$$$% %%%%%%&&D&N&&&&& ' 'I'T'T''o(p(j)u)u)*****8*C**********>+?+@+V+W+W+X+X+Y+Y+`+a+b+c+d+d+e+e+++(,),*,,,-----,-.-9-B-C-M-n-o-y-y-------///// /7/>/?/?/A/1:1111L2X2Y2445p6p6667;%D&D'DGKULZTkW\\\\b]__2`:`bbc5c8c M(:,T6^`OJPJQJ^Jo( ^`OJQJo(o pp^p`OJQJo( @ @ ^@ `OJQJo( ^`OJQJo(o ^`OJQJo( ^`OJQJo( ^`OJQJo(o PP^P`OJQJo(^`OJPJQJ^Jo(-^`OJQJ^Jo(hHop^p`OJQJo(hH@ ^@ `OJQJo(hH^`OJQJ^Jo(hHo^`OJQJo(hH^`OJQJo(hH^`OJQJ^Jo(hHoP^P`OJQJo(hH MT         $L        - >KO aE"0@[|Na OB >KO b)"YQCSiR b)9b w?CSiZUtV!)0")CSi*4+E"0}5CSim^6oD7CSi9R 2;E"0/mFE"0NzJR^L4LrBO3hR`T dG7|UE"0H@V\ZV*]pmd dE"0lTfCSihCSi(i d4rCUsCSi^w %V4aj@.I2Gdox*],~,T.A29FVp%,050stunx*-e16:OVY\\N{ !'V>\fgy 5?Mg`iDt}H&&*)55yAQz[-` egis{[F\wp'  n ' A M Q b n$ b %c g j s x &  O@ {W ] *d h }  U2 R S \ c h {  z  ? G W _ }  q \AeB~LL]e_Yv&[|bx!&vO\hml|'*Q;H<FYPYJly&  T0kM_Aav!h3Dj|:c#)_jh+3<F$L WfL[ <'[bcnvx&8c!;12S5apX~#&DJxMN]t"(=JCN R8U'x}='FHK\`i8~ 6R 7BIZ\`]!*4{6y< 7 tP<HClwyxh;c<p<\Oq4   : G= wE [ x !O !!_!_!e!x!""9"G"l"Qs"v"K###0#e#,B,\,g, -1-2-4-;-R-_-a-d-/g-4.%.r'.$1.1.F.Q.BW.jo.x.|./p'/w3/za/w000000'0C40K0[v0z01!1(1NR1Y1~2<2T2f2r2 33'3L/343:3R3 d39i3 4 44D4`P4m4D5H56m646*06G6L6O6O6px66% 7H7/747wE7g7n7&8sP8V8Sy89J9^9.d9Il9{9:D:b:5:9:d:q:p; ;,;5;J;K;O;fZ;v;<<|_<J`<c<d<j<m<v{<{<}<(=!=6=Q=R=,=>> >">(>?8 ??!?!>?MV?i?v?@@_6@I@Q@Q@`o@=}@AA9 A#>A8`AkAnA-vA|B $B4B5B\B`BqBrBCCCCC9COCC{CCLCeCD1D:DHDiPDfDE.E0EFE\ELFFFCF3uFFzG,G5G6sGHH7"HE=H?HGHNPH*_HqHitH~H3IAI$QIfaIlI8J JJ#Ja+J5Js[JpJtJH|Jb}JK KK*Ko0K[vKwKxKxK%L L#L&L:LYYLsLyL9{L M6MA;MNAMAMCMJEMOMbMeMfMoMoMvM2}MN#&N&NXNkNBOO:OGOVODtOuOuO P!P&P1P$6P9P?P?PdPVlP"Q17Q%ZQ:ZQ!rQ!yQR(R 0R7RZAR_GR_RmRqRQ;SGS=uSTd>TCTdVTVTo{T}T}T U U4U!U$U81U\UQ^UeUoU@VVGVLVNVqPV^VGWGW(W2W3WVOW]WaWfWsWXl X*X0X5XVXX]XcXmX^yXYY YJ#Y;YcY|Y Z#'Z,Z,ZM:Z[ <[<[H[Z[%f[i[{\$2\7\D\D\VU\_v\O]])/]Y6]pO]S]Ka]d]j]Ul]^^+^1^I=^n^{^|^ __#_X$_d_\r_`W&`8`c@`{G`P`R`^`c`v`{`Ma aJ0a?abrbb g#g=Bg9lgrgzgT)h1hDhEhVhXhbh#}h i> i&iv'iT8iDiMiiijj&j8(jj*jBjEj1]jyjQ0k23kEJknkl,l5lJlVl\l.mMmh#mW&mW+m 9m>mQm?cmn# n nn@"n+n1nqMn[ncgnIvnooo0ox1o8o!Ao/ZodoFroCpFphqppspqqqDq.q1qAlq}rrr8r;r4[rur}{r5 ssdsAsKsTsws|s t2t;t[tttFuNuu&uGuRuju8kuvuvvv.v']viw'w?wPw{w xqxS{g<o)oFGOppp}K,^CbLP4_y(]DNRTx>NLXb e~&29NT,[}Q~ #[F k&**>h|s.>jIRKLUd.&L}Y3+`%bs"||6v` M*68rfBNPu vd&3!>W0Z8TU;W=`S_Ghj|}~^01\35!#$56E Q/Z __}\ %+6fOzIM1hIIRtJj.8L h%#BMqQ[hl'q"y -7CFJ;Pfl1EHFIQcy})8ip}#~M<7 .'Y{Y-7DKMXnOo'!$*2<FcIcmb$3 CXGqv*1l}  )7M+cv-:Quuj-%383{VXEoQ<7=HCU?krV2PXoNv z&._7?` b i<?NR<^a(hhn N'+!D*&Q8BR JYqr"*,3VYhAqq$q@HV>bow/!`*KRbm %z 3<LiBqr".FGI\b "..>Vju~+-/0<OrW"ewP .MJX%dj9lov|! %26HB$RS|{}}a 4?erh]1diky{~% h$6r7;>V`be{ <%;^@7VCrm%a+~ou!@,?zV TYa_5g@IS XZ@RRUncG][e'+y/9<R5ej8q "N""@2Flz()15mu-8IlX x!%]^'!^ %-6Eayw"<E)FT`l[_c1ksA7AEWoW DKVcfy9+JU'w y}K+|.9PXks ][&k'_6C7xM#|?@M_*/BLPDliltxi-6A6>J5[^_Jby{56:m=SVdt5}a 83h<yJ` :!\q2B[\``Aaaeq'r4x6c8c__Grammarly_42____i__Grammarly_42___14H4sIAAAAAAAEAKtWckksSQxILCpxzi/NK1GyMqwFAAEhoTITAAAAXH4sIAAAAAAAEAKtWcslP9kxRslIyNDYyMDQ0NTWwMDI0MDUxNjZU0lEKTi0uzszPAykwNDauBQDMbpS/LgAAAA==@55556c@UnknownG.[x Times New Roman5Symbol3. .[x Arial;Wingdings9. . Segoe UI?= .Cx Courier NewA$BCambria Math"1hUXTDbhT 2hT 2!4cc2QHP ?cY2!xx Database SystemsWilliam Shoaff Debasis Mitra  Oh+'0  < H T `lt|Database SystemsWilliam Shoaff Normal.dotmDebasis Mitra15Microsoft Office Word@b@=@V@#R' hT ՜.+,D՜.+,H hp  Florida Tech2c Database Systems Title 8@ _PID_HLINKSAH$vL6https://www.cs.ucr.edu/~eamonn/time_series_data_2018/  1https://en.wikipedia.org/wiki/K-means_clustering^ 5https://www.ics.uci.edu/~alspaugh/cls/shr/allen.html\U&http://www.cs.fit.edu/~dmitra/ArtInt/\U&http://www.cs.fit.edu/~dmitra/ArtInt/rWmailto:dmitra@zach. fit.edu  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entry F`UFX'Data 1TablexYWordDocument .>SummaryInformation(DocumentSummaryInformation8MsoDataStoreB3X'B3X'HJTPE4MP5C==2B3X'B3X'Item  PropertiesUCompObj r   F Microsoft Word 97-2003 Document MSWordDocWord.Document.89q