Keynote at IEEE Services 2021: Abstract: https://conferences.computer.org/services/2021/keynotes/sheth.html Video: https://lnkd.in/d-r3YXaC Video of the same keynote given at DEXA2021: https://www.youtube.com/watch?v=u-06kK9TysA September 9, 2021 15:00 - 16:20 UTC ABSTRACT Knowledge representation as expert system rules or using frames and a variety of logics played a key role in capturing explicit knowledge during the hay days of AI in the past century. Such knowledge, aligned with planning and reasoning is part of what we refer to as Symbolic AI. The resurgent AI of this century in the form of Statistical AI has benefitted from massive data and computing. On some tasks, deep learning methods have even exceeded human performance levels. This gave the false sense that data alone is enough, and explicit knowledge is not needed. But as we start chasing machine intelligence that is comparable with human intelligence, there is an increasing realization that we cannot do without explicit knowledge. Neuroscience (role of long-term memory, strong interactions between different specialized regions of data on tasks such as multimodal sensing), cognitive science (bottom brain versus top brain, perception versus cognition), brain-inspired computing, behavioral economics (system 1 versus system 2), and other disciplines point to need for furthering AI to neuro-symbolic AI (i.e., hybrid of Statistical AI and Symbolic AI, also referred to as the third wave of AI). As we make this progress, the role of explicit knowledge becomes more evident. I will specifically look at our endeavor to support human-like intelligence, our desire for AI systems to interact with humans naturally, and our need to explain the path and reasons for AI systems’ workings. Nevertheless, the variety of knowledge needed to support understanding and intelligence is varied and complex. Using the example of progressing from NLP to NLU, I will demonstrate the dimensions of explicit knowledge, which may include, linguistic, language syntax, common sense, general (world model), specialized (e.g., geographic), and domain-specific (e.g., mental health) knowledge. I will also argue that despite this complexity, such knowledge can be scalability created and maintained (even dynamically or continually). Finally, I will describe our work on knowledge-infused learning as an example strategy for fusing statistical and symbolic AI in a variety of ways.

1. Don’t Handicap AI
Without Explicit
Knowledge
Keynote at IEEE Services 2021
Dr. Amit Sheth
Director of AI Institute
University of South Carolina
amit@sc.edu #AIISC, http://aiisc.ai

2. Big Data Is Enough for AI?
“Enough.”
Hinton: “Deep learning is going
to be able to do everything. ”
“Big data is enough. ”
“Not Really!”
Andrew Ng: “The importance of
big data is overhyped.”
Need higher levels of machine
intelligence
N. Gordon. “Don’t buy the ‘big data’ hype, says cofounder of Google Brain”, Fortune, July 2021.
2

3. No, Big Data Is Not Enough
P. Domingos. “A Few Useful Things to Know about Machine Learning”, Oct 2012.
J. Harris. “The Future of AI: Gary Marcus talks with Lex Fridman”, Medium, Oct 2019.
3
Pedro Domingos: “Data alone is not enough.” [2012]
Gary Marcus: “We need to build AI that captures how humans think. For example, analogy
researchers note that analogy is a ubiquitous component of human thinking”
Example of 2016 election: All data (all polls, all of social media, all of news articles) would
still not have predicted the correct outcome - a combination of data and human insights
(time, geography, demographics) allowed the correct outcome prediction.
Humans do not rely on data alone!

4. 2010
Paper “The Unreasonable Effectiveness of Data”:
● For many tasks, words and word combinations
provide all the representational machinery we need
to learn from text.
● So, follow the data. Choose a representation that
can use unsupervised learning on unlabeled data,
which is so much more plentiful than labeled data.
Represent all the data with a nonparametric model
rather than trying to summarize it with a parametric
model, because with very large data sources, the
data holds a lot of detail.
Now
Gary Marcus:
● “If you want to build a neural model of how humans do
certain class of things you’re going to have to change the
architecture”
● The first thing we have to do is to replace deep learning
with deep understanding. So you can’t have alignment with
a system that traffics only in correlations and doesn’t
understand concepts like “bottles” or “harm”
● We need to build AI that captures how humans think For
example, analogy researchers note that analogy is a
ubiquitous component of human thinking
● I think Hinton’s just wrong. Hinton says “we don’t want
hybrids”. People work towards hybrids and they will relabel
they’re hybrids is deep learning
● If you have a perceptual classification problem, throwing a
lot of data at it is better than anything else. But that has
not given us any material progress in natural language
understanding, common sense reasoning
Further Reading
4
Background slide

5. Focus of Most AI Systems
Classification Recommendation
Prediction Language Processing and Text Generation
What else do we need for higher levels of
machine intelligence?
5

6. https://www.istockphoto.com/photo/photophobia-or-eyes-sting-gm858772962-141858987
https://blog.kevineikenberry.com/leadership-supervisory-skills/the-best-leaders-carry-a-flashlight/
Narrow, well-defined tasks
(Reflects lower-levels of human-like
intelligence)
Human-like, broad spectrum behavior for
“looking after humans, companion to humans”
(Reflects higher-levels of human-like intelligence:
broad, complex, multi-faceted)
6

7. Abstraction
Contextualization
Personalization
Analogy
Causality
Key Higher-Level Capabilities
7

8. What constitutes human decision-making and communication
Multi-faceted, not just a well-defined script
We need to have a good model of the human
Mimicking Human Intelligence
What is human intelligence
Brain exploits both - perception (highly statistical in nature) AND cognition
(much is based on explicit structured knowledge)
A. Sheth, P. Anantharam, C. Henson. “Semantic, Cognitive, and Perceptual Computing: Paradigms That Shape Human Experience”, 2016.
8
Background slide

9. Theory of Multiple Intelligences
Triarchic Theory of Intelligence
Types of Human Intelligence
9
Background slide

10. Subbarao Kambhampati (Rao)
Recent advances have made AI synonymous with learning from massive amounts of
data, even in tasks for which we do have explicit theories and hard-won causal
knowledge
It is important for AI systems to be able to take knowledge when it is readily
available, rather than insist on rediscovering it indirectly from examples and
observation
The current zeal to spurn hard-won explicit (and often causal) knowledge, only to try
to (re)learn it from examples and traces as tacit knowledge, is quixotic at best
Humans have managed to develop shared representations and ability to
communicate via explicit knowledge, AI systems based purely on learning may well be
able to get there eventually
When our systems learn their own representations from raw data, there is little
reason to believe that their reasoning will be interpretable to us in any meaningful
way
AI’s romance with tacit knowledge has obvious adverse implications to safety,
correctness, and bias of our systems
Many of the pressing problems being faced in the deployment of AI technology,
including the interpretability concerns, the dataset bias concerns as well as the
robustness concerns can be traced rather directly back to the singular focus on
learning tacit knowledge from data, unsullied by any explicit knowledge taken from
the humans
Amit Sheth
Hinton is actually moving towards
explicit knowledge representation
in “representing part-whole
hierarchies in the neural network”
Two decades of using knowledge
graph in semantic search
Bottom up (“system 1”) and top-
down (“system 2”) has long been
advocated
Humans need to reach an
agreement on AI system’s
representations in order to
achieve interpretability and
explainability
It is impossible to discuss an
agreement with tacit
representations.
It is practical to build knowledge
graphs and the cost of building
them is justified
Geoffrey Hinton
“Deep learning is going
to be able to do
everything”
With a few conceptual
breakthroughs, deep
learning will be able to
replicate all of human
intelligence
Particularly
breakthroughs to do
with how to get big
vectors of neural activity
to implement things like
reason
Also need a massive
increase in scale because
the brain has about 100
trillion parameters
(synapses)
Further Reading
10
Background slide
Hinton Rao Sheth

11. Changing Expectations For AI
Deep learning methods work for narrowly defined tasks
This gives a false sense that data alone is enough
Need better forms of machine intelligence
From pattern recognition on massive datasets to higher levels of intelligence
Knowledge is needed to achieve human-level intelligence
Big data is no longer enough, explicit knowledge is required
11

12. Changing Expectations For AI
Mimicking human intelligence
Reasoning, Analogy, Spatio-temporal, Causality, Biases, etc.
Understanding the data in the context of serving human needs
Personalization, Contextualization, and Abstraction
Knowledge needs to play equally important roles
For higher level intelligence, both Knowledge and data has to play an important
role
12

13. STATISTICAL
AI
CONNECTIONIST
“Unreasonable effectiveness
of big data”
in machine processing &
powering bottom up processing
“Unreasonable effectiveness of
small data”
in human decision making - can
this be emulated to power top
down processing?
SYMBOLIC AI
FORMAL
KG will play an increasing role in developing hybrid neuro-symbolic systems (that is bottom-up deep
learning with top-down symbolic computing) as well as in building explainable AI systems for which KGs
will provide scaffolding for punctuating neural computing.
Cognitive Science Analogy: Combining Top Brain - Bottom Brain Processes.
Statistical v.s. Symbolic AI
13
Background slide

14. Symbolic vs Statistical AI
Knowledge representation as expert system rules or using frames and
variety of logics, played a key role in capturing explicit knowledge during
the hay days of AI in the past century
Such knowledge, aligned with planning and reasoning are part of what we
refer to as Symbolic AI
The resurgent AI of this century in the form of Statistical AI has benefitted
from massive data and computing
Statistical AI is NOT intelligence
14
Background slide

15. [Explicit] Knowledge
will play a central role
15
Sheth, Thirunarayanan: The Duality of Data and Knowledge Across the Three Waves of AI, 2021

16. From Statistical to Causal & Explainable AI
We need to go beyond systems that merely get better and better at
detecting statistical patterns in datasets
Start building AI systems that from the moment of their assembly
innately grasp basic concepts:
❏ Time, Space
❏ Knowledge and Experiences
❏ Causality
16

17. Statistical AI Alone Cannot Achieve Full Intelligence
Humans take for granted common sense knowledge when expressing natural language
❏ For AI, identifying the implicit common sense to capture full context and meaning is not possible
from statistical pattern matching on the apparent text! (NLP)
❏ Modeling the common sense through external knowledge is essential (NLU)
❏ This concern is not limited to language understanding
17
?

18. Knowledge and
Experience
System 1
Perception
Statistical AI
System 2
Cognition
Symbolic AI
Data to Concepts
and
Abstractions
System 1 (Statistical) & System 2 (Symbolic)
18

19. 19
A. Sheth, P. Anantharam, C. Henson. “Semantic, Cognitive, and Perceptual Computing: Paradigms That Shape Human Experience”, 2016. Keynote talk

20. www.presentationgo.com
Explicit Knowledge
● Can be readily articulated, codified, stored
and accessed
● Easy to transfer to others
● Can be stored in certain media
Tacit Knowledge
● Not codified
● Difficult to express or extract
● Hard to transfer to others by writing down or
verbalizing
● Can include personal wisdom, experience,
insight, intuition
20

21. 21
Humans use
knowledge
In all types of decision,
activities and actions.
We will need to model
and use knowledge in
AI systems that assist
in or undertake these
functions.

22. Converts processing into understanding
E.g. Natural language extraction (processing) vs. Natural language
understanding (must have knowledge)
The Roles of Knowledge
Improves deep learning
May require less data
Identifies biases in data
Only using data can be misleading
22
Background slide

23. Knowledge Representation
“There is no machine intelligence without knowledge representation.”- Adrian Bowles
KGs will play an increasing role in developing hybrid neuro-symbolic systems (that is
bottom-up deep learning with top-down symbolic computing) as well as in building
explainable AI systems for which KGs will provide scaffolding for punctuating neural
computing
It is possible to have these and other features that could enable humans to trust an
AI system only if it is possible for humans to reach an agreement among them
M. Knight. “Taxonomy vs Ontology: Machine Learning Breakthroughs”, DATAVERSITY, Oct 2017.
A. Sheth. “Mini commentary and reflection on Rao's article on Romance with Tacit Knowledge”, Feb 2021.
23
Background slide

24. Structured knowledge can capture and represent the full richness
associated with human intelligence
Knowledge constructs: the ability to capture declarative knowledge, encode
abstract notions, causal and predictive models
Rely on explicit concepts and well-identified, overtly defined relations
rather than machine embeddings in the latent space
Knowledge Is Essential for Higher
Machine Intelligence
G. Singer. “The Rise of Cognitive AI”, Towards Data Science, Apr 2021. 24
Background slide

25. Knowledge can capture common sense assumptions and the
underlying logic
Well-structured knowledge representation can address aspects
of disambiguation
More extensive potential for explanations and predictions well
beyond the capabilities of a statistical mapping function
25
G. Singer. “The Rise of Cognitive AI”, Towards Data Science, Apr 2021.
Knowledge Is Essential for Higher
Machine Intelligence

26. Machine
Intelligence:
From Data to
Decisions
26
Background slide

27. Three technical reasons why statistical approaches fail for NLU
Progressing From NLP to NLU
NLP is a completely different problem from NLU
1. Missing Text Phenomenon (What references are implicit?)
2. Intension (Why did they express the language that way?)
3. Statistical (In)Significance (Ex: The main entity appears seldom in
the passage due to pronouns - Xanadu appears once!)
Conclusion: Language is not just data
W. Saba. “Time to put an end to BERTology (or, ML/DL is not even relevant to NLU)”, Medium, Oct 2020.
27

28. NLP vs. NLU
28

29. KILU Tasks to Test NLU vs. GLUE Tasks to Test NLP
Knowledge Intensive Language Understanding (KILU)
General Language Understanding Evaluation (GLUE)
A.Sheth, M. Gaur, K. Roy, K.Faldu. "Knowledge-intensive Language Understanding for Explainable AI," IEEE Internet Computing, September/October 2021.
29

30. Without Domain Knowledge
With Domain Knowledge
KILU Tasks - From NLP to NLU
Understanding a drug
31

31. Without
Contextualized Personalization
With
Contextualized Personalization
KILU Tasks - From NLP to NLU
Asthma Virtual Assistant
32

32. I’ve had a long day at work and
I would like to go grab a drink
Sure, there is a bar at […..]
that’s got great reviews
Alright, thank
you.
I’ve had a long day at work and
I would like to go grab a drink
Your depression typically gets
worse with alcohol, are you sure?
You’re right.
Thanks
Without Personalization With Personalization
Adam Alcoholism
Suffers from
PKG:
Personalized knowledge graph
(PKG) support the wisdom to
prevent Adam from going out to
a bar, which may be a norm with
extenuating circumstances
Personalized Virtual Health Assistant
33
Background slide

33. KILU Tasks - From NLP to NLU
Mental Health Virtual Assistant
In this task, natural language
generation requires patient and
domain expert language
understanding through:
1. Personalization: by
tracking patient profile as
knowledge
2. Abstraction: mapping
patient features to human
understandable concepts
3. Continuous
contextualization: Which
part of the medical
knowledge is applicable?
34

34. Tasks Outside Natural Language Domain That
Require Better Intelligence
Self-Driving Cars
Ethics, law, priorities...
Conversations With a Patient
Empathy, past history, continuity, desired outcome...
Effective Educational Pedagogy
Student preparation, analogy, desired outcomes...
Supporting Social-Good
E.g., countering anti-vax messages
(understanding of motivation of a group of advocates)
35

35. 36
Driving Domain - Traffic Flow Analysis:
Multiple knowledge graphs enabled connection of data of diverse modality
36
Anantharam, et al. Understanding City Traffic Dynamics Utilizing Sensor and Textual Observations. 2016.

36. Analogy-making can give us the conceptual understanding needed
for abstraction. And abstraction is what gives the humans the ability
to generalize from situations that he/she have not seen before to
situations that they know. The brittleness of AI systems results at
least partly because it cannot generalize from few examples (need
thousands/ millions of data points to “learn” a single concept
whereas humans can generalize from situations that they have
never seen before).
38
Above is not a quote from - but see for related discussion Abstraction and Analogy-Making in Artificial Intelligence by Melanie Mitchell

37. Without Knowledge We Can’t Climb
the Abstraction Ladder
Personification
I wandered lonely
as a cloud
Poetry
Art
Creativity
Apple
Granny Smith
Fruit
Low-calorie
Health
39
https://www.rijnlandmodel.nl/achtergrond/algemene_semantiek/hayakawa/ch10_abstraction-ladder.gif

38. Analogical Understanding
41
Developed by Prof. Nitin Jain at U Tennessee

39. It is hard to represent Analogy
without explicit knowledge
Abstraction and mapping to analogical concepts that form the sequence
of steps in the process is impossible without Knowledge Graphs
Analogical reasoning requires representing systematicity and process
constrained sequential process
Whitaker et al. “Neuroscientific insights into the development of analogical reasoning”, Developmental Science, 2017.
A. Gomila. “6 - Language as Tool Kit, 1: Representational Effects”, Verbal Minds, 2012. 42

40. ❏ Rope = electron transport chain
❏ Villager pulling rope provides
energy = electron transfer on
transport chain provides energy
❏ Draws water from well to flow down
the slide due to potential gradient =
draws protons from mitochondria
matrix flow across the membrane
due to proton gradient.
❏ Water on wheel generates motor
force, rotating the wheel, to
generate kinetic energy = proton
flow generates proton motor force,
rotating c-subunits and gamma-
subunits of ATP synthase, to
generate ATP (kinetic energy)
Analogical Understanding
43

41. High-Level Knowledge
■ Analogical reasoning: the ability to compare
with past experience
■ Process knowledge: reason using valid
actions dictated by their process knowledge
obtained during life experiences
○ Cognitive processes - postulates mental
processes
○ Behavioral processes - cause-effect
relationships without mental
processing
■ Planning: reasoning over the knowledge to
enable actions
■ Representation (abstraction) - mediated
cognition: reasoning with representations of
things to person has experienced (e.g., valid
actions as allowed by physics)
44

42. Think
Abstract
Act
Decide
Compiling relevant knowledge for abstraction and
reasoning based on task and observational data
Abstracting observation data by mapping to compiled knowledge
(during thinking) to form human understandable concepts
Synthesizing/reasoning over the concepts using relevant knowledge
compiled during the thinking phase resulting in action
Executing the action to enable decision
The Roles of Knowledge
45

43. Interpretability
ML/DL
Model
ML/DL
Model
Knowledge
Graph Recursive
Explainability
Interpretability+Traceability
46

44. Machine Learning models provide interpretability through parameter
visualization methods (highlighting phrases)
Knowledge Is Need for
Explainability
47
Knowledge is necessary to identify phrases that pattern recognition cannot
detect
Knowledge traceability provides explanations on the model outcome in terms
of concepts that the domain experts can understand.
Semantics of the Black-Box: Can Knowledge Graphs Help Make Deep Learning Systems More Interpretable and Explainable?

45. 48
Approaches to
Incorporating/Infusing Knowledge
in Deep Learning
Sheth et al, Shades of Knowledge-Infused Learning for Enhancing Deep Learning, 2019

46. Dataset
Enrich
Deep Learning Model
Tacit
Knowledge
Hypothesis
Testing or
Similarity-Based
Verification
Shallow Infusion
Tacit
Knowledge
Self-Aware or
External Knowledge Self-Aware or
External Knowledge
Similarity
Based
Verification
Semi-Deep Infusion
Dataset
Deep Learning Model
Architecture For Knowledge Infused Learning
49

47. Really struggling with my bisexuality which is causing chaos in my relationship with a girl.
Being a fan of LGBTQ community, I am equal to worthless for her. I’m now starting to get
drunk because I can’t cope with the obsessive, intrusive thoughts, and need to get it out of
my head.
Is mental health related ? Yes: 0.71 , No: 0.29
Which mental health condition?
Predicted: Depression (False)
True: Obsessive Compulsive Disorder
Reasoning over Model:
Why model predicted
Depression?
Unknown
Manas Gaur, Ugur Kursuncu, Amanuel Alambo, Amit Sheth,
Raminta Daniulaityte, Krishnaprasad Thirunarayan, and
Jyotishman Pathak. "" Let Me Tell You About Your Mental
Health!" Contextualized Classification of Reddit Posts to DSM-
5 for Web-based Intervention." CIKM 2018. [CIKM 2018]
Data-Driven Reasoning Without Knowledge
50
Highlighted terms
based on attention
matrix

48. Really struggling with my bisexuality which is causing chaos in my relationship with a girl.
Being a fan of LGBTQ community, I am equal to worthless for her. I’m now starting to get
drunk because I can’t cope with the obsessive intrusive thoughts, and need to get it out of
my head.
Is mental health related ? Yes: 0.82 , No: 0.18
Which mental health condition?
Predicted: Obsessive Compulsive Disorder(True)
True: Obsessive Compulsive Disorder
DSM-5 Knowledge
Graph
DSM-5 and Post
Correlation Matrix
Reasoning over model:
Why model predicted
Obsessive Compulsive
Disorder ? known
Interpretable Learning
D
εRN
P εRN
W f(W)
Data-Driven Reasoning With Knowledge
51

49. Really struggling with my [health-related behavior] which is causing [health-related
behavior] with a girl. Being a fan of [XXX] community, I am equal to [level of mood] for
her. I’m now starting to [drinking] because I can’t cope with the [obsessive compulsive
personality disorder] [disturbance in thinking], and [disturbance in thinking].
Is mental health related ? Yes: 0.96 , No: 0.04
Which Mental Health condition?
Predicted: Obsessive Compulsive Disorder(True)
True: Obsessive Compulsive Disorder
DSM-5 Knowledge
Graph
DSM-5 and Post
Correlation Matrix
Reasoning over Model:
Why model predicted
Obsessive Compulsive
Disorder ? known
Interpretable and
Explainable Learning
Semantic Annotation
52

50. Really struggling with my bisexuality which
is causing chaos in my relationship with a
girl. I am equal to worthless for her. I’m now
starting to get drunk because I can’t cope
with the obsessive, intrusive thoughts, and
need to get out of my head.
288291000119102: High risk bisexual behavior
365949003: Health-related behavior finding 365949003: Health-related behavior finding
307077003: Feeling hopeless
365107007: level of mood
225445003: Intrusive thoughts
55956009: Disturbance in content of thought
26628009: Disturbance in thinking
1376001: Obsessive compulsive personality disorder
Multi-hop traversal on
medical knowledge
graphs
<is symptom>
Traceability of Concepts to the Knowledge
53
Obsessive-compulsive disorder is a disorder in
which people have obsessive, intrusive thoughts,
ideas or sensations that make them feel driven to do
something repetitively

51. K-IL: Semi-Deep Infusion : Matching Reddit Conversation to DSM-5
Domain-specific
Knowledge lowers
False Alarm Rates.
2005-2016
550K Users
8 Million
Conversations
15 Mental Health
Subreddits
[Gkotsis 2017]
[Saravia 2016]
[Park 2018]
Performance Gains in the outcomes
[CIKM 2018]
54
Gaur et al, "Let Me Tell You About Your Mental Health!": Contextualized Classification of Reddit Posts to DSM-5 for Web-based Intervention, 2018

52. Tacit
Knowledge
Similarity Based
Verification
Deep Infusion
Deep Learning Model
Stratified Layer 1 Stratified Layer 2
Stratified Knowledge 1 Stratified Knowledge 2
Architecture For Knowledge Infused Learning
55

53. Explicit knowledge will play a major role
early in the third wave.
Increasing machine intelligence so that
AI can engage and assist humans will
require learning from many disciplines.
NeuroSc Cognitive Sc Behavioral
Econ
Conclusions
56
DARPA Perspective on AI

54. Manas Gaur
Ph.D. Student
Artificial Intelligence
Institute, University of South
Carolina
Kaushik Roy
Ph.D. Student
Artificial Intelligence
Institute, University of South
Carolina
Acknowledgement
Yuxin Zi
Ph.D. Student
Artificial Intelligence
Institute, University of South
Carolina
Various NSF (e.g., “Advancing Neuro-symbolic AI with Deep Knowledge-infused Learning”
[NSF#2133842]) and NIH (e.g., AI/ML readiness for AI/ML-Readiness for Neuroimaging of
Language) grants. More at: http://wiki.aiisc.ai

55. [1] N. Gordon. “Don’t buy the ‘big data’ hype, says cofounder of Google Brain”, Fortune, July 2021.
[2] K. Hao. “AI pioneer Geoff Hinton: “Deep learning is going to be able to do everything””, MIT Technology Review, Nov 2020.
[3] P. Domingos. “A Few Useful Things to Know about Machine Learning”, Oct 2012.
[4] J. Harris. “The Future of AI: Gary Marcus talks with Lex Fridman”, Medium, Oct 2019.
[5] G. Marcus, E. Davis. “How to Build Artificial Intelligence We Can Trust”, The New York Times, Sep 2019
[6] W. Saba. Machine Learning Won't Solve Natural Language Understanding, The Gradient, Aug 2021.
[7] W. Saba. “Time to put an end to BERTology (or, ML/DL is not even relevant to NLU)”, Medium, Oct 2020.
[8] A.Sheth, M. Gaur, K. Roy, K.Faldu. "Knowledge-intensive Language Understanding for Explainable AI," IEEE Internet Computing, September/October 2021.
[9] A. Sheth, P. Anantharam, C. Henson. “Semantic, Cognitive, and Perceptual Computing: Paradigms That Shape Human Experience”, 2016.
[10] M. Knight. “Taxonomy vs Ontology: Machine Learning Breakthroughs”, DATAVERSITY, Oct 2017.
[11] A. Sheth. “Mini commentary and reflection on Rao's article on Romance with Tacit Knowledge”, Feb 2021.
[12] G. Singer. “The Rise of Cognitive AI”, Towards Data Science, Apr 2021.
[13] Whitaker et al. “Neuroscientific insights into the development of analogical reasoning”, Developmental Science, 2017.
[14] A. Gomila. “6 - Language as Tool Kit, 1: Representational Effects”, Verbal Minds, 2012.
References
58

56. 59
AI Institute’s Core and Translational/Interdisciplinary Research